# Licensed under a 3-clause BSD style license - see LICENSE.rst
"""
Tools for calculating the properties of sources defined by a
segmentation image.
"""
import functools
import inspect
import math
import warnings
from copy import deepcopy
import astropy.units as u
import numpy as np
from astropy.stats import SigmaClip, gaussian_fwhm_to_sigma
from astropy.utils import lazyproperty
from scipy.ndimage import map_coordinates
from scipy.optimize import root_scalar
from photutils.aperture import (BoundingBox, CircularAperture,
EllipticalAperture, RectangularAnnulus)
from photutils.background import SExtractorBackground
from photutils.geometry import circular_overlap_grid, elliptical_overlap_grid
from photutils.morphology import gini as gini_func
from photutils.segmentation.core import SegmentationImage
from photutils.segmentation.utils import _mask_to_mirrored_value
from photutils.utils._deprecation import (_get_future_column_names,
create_empty_deprecated_qtable,
deprecated_getattr,
deprecated_positional_kwargs,
deprecated_renamed_argument)
from photutils.utils._misc import _get_meta
from photutils.utils._progress_bars import add_progress_bar
from photutils.utils._quantity_helpers import process_quantities
from photutils.utils.cutouts import CutoutImage
__all__ = ['SourceCatalog']
# Default table columns for `to_table()` output
DEFAULT_COLUMNS = ['label', 'x_centroid', 'y_centroid', 'sky_centroid',
'bbox_xmin', 'bbox_xmax', 'bbox_ymin', 'bbox_ymax',
'area', 'semimajor_axis', 'semiminor_axis', 'orientation',
'eccentricity', 'min_value', 'max_value', 'segment_flux',
'segment_flux_err', 'kron_flux', 'kron_flux_err']
# Remove in 4.0
_DEPRECATED_ATTRIBUTES = {
'add_extra_property': 'add_property',
'apermask_method': 'aperture_mask_method',
'background': 'background_cutout',
'background_ma': 'background_cutout_masked',
'convdata': 'conv_data_cutout',
'convdata_ma': 'conv_data_cutout_masked',
'covar_sigx2': 'covariance_xx',
'covar_sigxy': 'covariance_xy',
'covar_sigy2': 'covariance_yy',
'cutout_maxval_index': 'cutout_max_value_index',
'cutout_minval_index': 'cutout_min_value_index',
'cxx': 'ellipse_cxx',
'cxy': 'ellipse_cxy',
'cyy': 'ellipse_cyy',
'data': 'data_cutout',
'data_ma': 'data_cutout_masked',
'error': 'error_cutout',
'error_ma': 'error_cutout_masked',
'extra_properties': 'custom_properties',
'fluxfrac_radius': 'flux_radius',
'get_label': 'select_label',
'get_labels': 'select_labels',
'kron_fluxerr': 'kron_flux_err',
'localbkg_width': 'local_bkg_width',
'maxval_index': 'max_value_index',
'maxval_xindex': 'max_value_xindex',
'maxval_yindex': 'max_value_yindex',
'minval_index': 'min_value_index',
'minval_xindex': 'min_value_xindex',
'minval_yindex': 'min_value_yindex',
'nlabels': 'n_labels',
'remove_extra_properties': 'remove_properties',
'remove_extra_property': 'remove_property',
'rename_extra_property': 'rename_property',
'segment': 'segment_cutout',
'segment_fluxerr': 'segment_flux_err',
'segment_ma': 'segment_cutout_masked',
'semimajor_sigma': 'semimajor_axis',
'semiminor_sigma': 'semiminor_axis',
'xcentroid': 'x_centroid',
'xcentroid_quad': 'x_centroid_quad',
'xcentroid_win': 'x_centroid_win',
'ycentroid': 'y_centroid',
'ycentroid_quad': 'y_centroid_quad',
'ycentroid_win': 'y_centroid_win',
}
_DEPRECATED_META_KEYS = {
'localbkg_width': 'local_bkg_width',
'apermask_method': 'aperture_mask_method',
}
def as_scalar(method):
"""
Return a decorated method where it will always return a scalar value
(instead of a length-1 tuple/list/array) if the class is scalar.
Note that lazyproperties that begin with '_' should not have this
decorator applied. Such properties are assumed to always be iterable
and when slicing (see __getitem__) from a cached multi-object
catalog to create a single-object catalog, they will no longer be
scalar.
Parameters
----------
method : function
The method to be decorated.
Returns
-------
decorator : function
The decorated method.
"""
@functools.wraps(method)
def _as_scalar(*args, **kwargs):
result = method(*args, **kwargs)
try:
return (result[0] if args[0].isscalar and len(result) == 1
else result)
except TypeError: # if result has no len
return result
return _as_scalar
def use_detcat(method):
"""
Return a decorated method where it will return the value from the
detection image catalog instead of using the method to calculate it.
Parameters
----------
method : function
The method to be decorated.
Returns
-------
decorator : function
The decorated method.
"""
@functools.wraps(method)
def _use_detcat(self, *args, **kwargs):
if self._detection_catalog is None:
return method(self, *args, **kwargs)
return getattr(self._detection_catalog, method.__name__)
return _use_detcat
[docs]
class SourceCatalog:
"""
Class to create a catalog of photometry and morphological properties
for sources defined by a segmentation image.
Parameters
----------
data : 2D `~numpy.ndarray` or `~astropy.units.Quantity`, optional
The 2D array from which to calculate the source photometry and
properties. If ``convolved_data`` is input, then a convolved
version of ``data`` will be used instead of ``data`` to
calculate the source centroid and morphological properties.
Source photometry is always measured from ``data``. For
accurate source properties and photometry, ``data`` should be
background-subtracted. Non-finite ``data`` values (NaN and inf)
are automatically masked.
segmentation_image : `~photutils.segmentation.SegmentationImage`
A `~photutils.segmentation.SegmentationImage` object defining
the sources.
convolved_data : 2D `~numpy.ndarray` or `~astropy.units.Quantity`, optional
The 2D array used to calculate the source centroid and
morphological properties. Typically, ``convolved_data``
should be the input ``data`` array convolved by the
same smoothing kernel that was applied to the detection
image when deriving the source segments (e.g., see
:func:`~photutils.segmentation.detect_sources`). If
``convolved_data`` is `None`, then the unconvolved ``data`` will
be used instead. Non-finite ``convolved_data`` values (NaN and
inf) are not automatically masked, unless they are at the same
position of non-finite values in the input ``data`` array. Such
pixels can be masked using the ``mask`` keyword.
error : 2D `~numpy.ndarray` or `~astropy.units.Quantity`, optional
The total error array corresponding to the input ``data``
array. ``error`` is assumed to include *all* sources of
error, including the Poisson error of the sources (see
`~photutils.utils.calc_total_error`). ``error`` must have
the same shape as the input ``data``. If ``data`` is a
`~astropy.units.Quantity` array then ``error`` must be a
`~astropy.units.Quantity` array (and vice versa) with identical
units. Non-finite ``error`` values (NaN and inf) are not
automatically masked, unless they are at the same position of
non-finite values in the input ``data`` array. Such pixels can
be masked using the ``mask`` keyword. See the Notes section
below for details on the error propagation.
mask : 2D `~numpy.ndarray` (bool), optional
A boolean mask with the same shape as ``data`` where a `True`
value indicates the corresponding element of ``data`` is masked.
Masked data are excluded from all calculations. Non-finite
values (NaN and inf) in the input ``data`` are automatically
masked.
background : float, 2D `~numpy.ndarray`, or `~astropy.units.Quantity`, \
optional
The background level that was *previously* present in the input
``data``. ``background`` may either be a scalar value or a 2D
image with the same shape as the input ``data``. If ``data``
is a `~astropy.units.Quantity` array then ``background`` must
be a `~astropy.units.Quantity` array (and vice versa) with
identical units. Inputing the ``background`` merely allows
for its properties to be measured within each source segment.
The input ``background`` does *not* get subtracted from the
input ``data``, which should already be background-subtracted.
Non-finite ``background`` values (NaN and inf) are not
automatically masked, unless they are at the same position of
non-finite values in the input ``data`` array. Such pixels can
be masked using the ``mask`` keyword.
wcs : WCS object or `None`, optional
A world coordinate system (WCS) transformation that
supports the `astropy shared interface for WCS
<https://docs.astropy.org/en/stable/wcs/wcsapi.html>`_ (e.g.,
`astropy.wcs.WCS`, `gwcs.wcs.WCS`). If `None`, then all
sky-based properties will be set to `None`. This keyword will be
ignored if ``detection_catalog`` is input.
local_bkg_width : int, optional
The width of the rectangular annulus used to compute a
local background around each source. If zero, then no local
background subtraction is performed. The local background
affects the ``min_value``, ``max_value``, ``segment_flux``,
``kron_flux``, and ``flux_radius`` properties. It is also used
when calculating circular and Kron aperture photometry (i.e.,
`circular_photometry` and `kron_photometry`). It does not affect
the moment-based morphological properties of the source.
aperture_mask_method : {'correct', 'mask', 'none'}, optional
The method used to handle neighboring sources when performing
aperture photometry (e.g., circular apertures or elliptical Kron
apertures). This parameter also affects the Kron radius. The
options are:
* 'correct': replace pixels assigned to neighboring sources by
replacing them with pixels on the opposite side of the source
center (equivalent to MASK_TYPE=CORRECT in SourceExtractor).
* 'mask': mask pixels assigned to neighboring sources
(equivalent to MASK_TYPE=BLANK in SourceExtractor).
* 'none': do not mask any pixels (equivalent to MASK_TYPE=NONE
in SourceExtractor).
This keyword will be ignored if ``detection_catalog`` is
input. In that case, the ``aperture_mask_method`` set in the
``detection_catalog`` will be used.
kron_params : tuple of 2 or 3 floats, optional
A list of parameters used to determine the Kron aperture.
The first item is the scaling parameter of the unscaled Kron
radius and the second item represents the minimum value for
the unscaled Kron radius in pixels. The optional third item is
the minimum circular radius in pixels. If ``kron_params[0]``
* `kron_radius` * sqrt(`semimajor_axis` * `semiminor_axis`)
is less than or equal to this radius, then the Kron aperture
will be a circle with this minimum radius. This keyword will be
ignored if ``detection_catalog`` is input.
detection_catalog : `SourceCatalog`, optional
A `SourceCatalog` object for the detection image. The
segmentation image used to create the detection catalog must be
the same one input to ``segmentation_image``. If input, then
the detection catalog source centroids and morphological/shape
properties will be returned instead of calculating them from
the input ``data``. The detection catalog centroids and shape
properties will also be used to perform aperture photometry
(i.e., circular and Kron). If ``detection_catalog`` is
input, then the input ``wcs``, ``aperture_mask_method``, and
``kron_params`` keywords will be ignored. This keyword affects
`circular_photometry` (including returned apertures), all Kron
parameters (Kron radius, flux, flux errors, apertures, and
custom `kron_photometry`), and `flux_radius` (which is based on
the Kron flux).
progress_bar : bool, optional
Whether to display a progress bar when calculating
some properties (e.g., ``kron_radius``, ``kron_flux``,
``flux_radius``, ``circular_photometry``, ``centroid_win``,
``centroid_quad``). The progress bar requires that the `tqdm
<https://tqdm.github.io/>`_ optional dependency be installed.
Notes
-----
``data`` should be background-subtracted for accurate source
photometry and properties. The previously-subtracted background can
be passed into this class to calculate properties of the background
for each source.
Note that this class does not convert input data in
surface-brightness units to flux or counts. Conversion from
surface-brightness units should be performed before using this
class.
`SourceExtractor`_'s centroid and morphological parameters are
always calculated from a convolved, or filtered, "detection"
image (``convolved_data``), i.e., the image used to define the
segmentation image. The usual downside of the filtering is the
sources will be made more circular than they actually are. If
you wish to reproduce `SourceExtractor`_ centroid and morphology
results, then input the ``convolved_data``. If ``convolved_data``
is `None`, then the unfiltered ``data`` will be used for the source
centroid and morphological parameters.
Negative data values within the source segment are set to zero
when calculating morphological properties based on image moments.
Negative values could occur, for example, if the segmentation
image was defined from a different image (e.g., different
bandpass) or if the background was oversubtracted. However,
`~photutils.segmentation.SourceCatalog.segment_flux` always includes
the contribution of negative ``data`` values.
The input ``error`` array is assumed to include *all* sources
of error, including the Poisson error of the sources.
`~photutils.segmentation.SourceCatalog.segment_flux_err` is simply
the quadrature sum of the pixel-wise total errors over the
unmasked pixels within the source segment:
.. math::
\\Delta F = \\sqrt{\\sum_{i \\in S}
\\sigma_{\\mathrm{tot}, i}^2}
where :math:`\\Delta F` is
`~photutils.segmentation.SourceCatalog.segment_flux_err`,
:math:`S` are the unmasked pixels in the source segment, and
:math:`\\sigma_{\\mathrm{tot}, i}` is the input ``error`` array.
Custom errors for source segments can be calculated using the
`~photutils.segmentation.SourceCatalog.error_cutout_masked` and
`~photutils.segmentation.SourceCatalog.background_cutout_masked`
properties, which are 2D `~numpy.ma.MaskedArray` cutout versions of
the input ``error`` and ``background`` arrays. The mask is `True`
for pixels outside the source segment, masked pixels from the
``mask`` input, or any non-finite ``data`` values (NaN and inf).
**Scalar vs. Multi-source Catalogs**
A `SourceCatalog` can represent a single source or multiple
sources. Most properties adapt their return type accordingly: for
a multi-source catalog, properties return arrays or lists (one
element per source); for a single-source (scalar) catalog, the
same properties return a scalar value or a single object. For
example, `kron_aperture` returns a list of aperture objects for a
multi-source catalog, but a single aperture object for a scalar
catalog. Similarly, `data_cutout` returns a list of 2D cutout arrays
for a multi-source catalog, but a single 2D array for a scalar
catalog. A scalar catalog is created when the a multi-source catalog
is indexed to select a single source.
.. _SourceExtractor: https://sextractor.readthedocs.io/en/latest/
"""
@deprecated_renamed_argument('segment_img', 'segmentation_image', '3.0',
until='4.0')
@deprecated_renamed_argument('localbkg_width', 'local_bkg_width',
'3.0', until='4.0')
@deprecated_renamed_argument('apermask_method', 'aperture_mask_method',
'3.0', until='4.0')
@deprecated_renamed_argument('detection_cat', 'detection_catalog', '3.0',
until='4.0')
def __init__(self, data, segmentation_image, *, convolved_data=None,
error=None, mask=None, background=None, wcs=None,
local_bkg_width=0, aperture_mask_method='correct',
kron_params=(2.5, 1.4, 0.0), detection_catalog=None,
progress_bar=False):
inputs = (data, convolved_data, error, background)
names = ('data', 'convolved_data', 'error', 'background')
inputs, unit = process_quantities(inputs, names)
(data, convolved_data, error, background) = inputs
self._data_unit = unit
self._data = self._validate_array(data, 'data', shape=False)
self._convolved_data = self._validate_array(convolved_data,
'convolved_data')
self._segmentation_image = self._validate_segmentation_image(
segmentation_image)
self._error = self._validate_array(error, 'error')
self._mask = self._validate_array(mask, 'mask')
self._background = self._validate_array(background, 'background')
self.wcs = wcs
self.local_bkg_width = self._validate_local_bkg_width(
local_bkg_width)
self.aperture_mask_method = self._validate_aperture_mask_method(
aperture_mask_method)
self.kron_params = self._validate_kron_params(kron_params)
self.progress_bar = progress_bar
# Needed for ordering and isscalar
# NOTE: calculate slices before labels for performance.
# _labels is initially always a non-scalar array, but
# it can become a numpy scalar after indexing/slicing.
self._slices = self._segmentation_image.slices
self._labels = self._segmentation_image.labels
if self._labels.shape == (0,):
msg = 'segmentation_image must have at least one non-zero label'
raise ValueError(msg)
self._detection_catalog = self._validate_detection_catalog(
detection_catalog)
attrs = ('wcs', 'aperture_mask_method', 'kron_params')
if self._detection_catalog is not None:
for attr in attrs:
setattr(self, attr, getattr(self._detection_catalog, attr))
if convolved_data is None:
self._convolved_data = self._data
self._aperture_mask_kwargs = {
'circ': {'method': 'exact'},
'kron': {'method': 'exact'},
'flux_radius': {'method': 'exact'},
'cen_win': {'method': 'center'},
}
self.default_columns = DEFAULT_COLUMNS
self._custom_properties = []
self._flux_radius_cache = {}
self.meta = _get_meta()
self._update_meta()
def _validate_segmentation_image(self, segmentation_image):
if not isinstance(segmentation_image, SegmentationImage):
msg = 'segmentation_image must be a SegmentationImage'
raise TypeError(msg)
if segmentation_image.shape != self._data.shape:
msg = 'segmentation_image and data must have the same shape'
raise ValueError(msg)
return segmentation_image
def _validate_array(self, array, name, *, shape=True):
if name == 'mask' and array is np.ma.nomask:
array = None
if array is not None:
# UFuncTypeError is raised when subtracting float
# local_background from int data; convert to float
array = np.asanyarray(array)
if array.ndim != 2:
msg = f'{name} must be a 2D array'
raise ValueError(msg)
if shape and array.shape != self._data.shape:
msg = f'data and {name} must have the same shape'
raise ValueError(msg)
return array
@staticmethod
def _validate_local_bkg_width(local_bkg_width):
if local_bkg_width < 0:
msg = 'local_bkg_width must be >= 0'
raise ValueError(msg)
local_bkg_width_int = int(local_bkg_width)
if local_bkg_width_int != local_bkg_width:
msg = 'local_bkg_width must be an integer'
raise ValueError(msg)
return local_bkg_width_int
@staticmethod
def _validate_aperture_mask_method(aperture_mask_method):
if aperture_mask_method not in ('none', 'mask', 'correct'):
msg = 'Invalid aperture_mask_method value'
raise ValueError(msg)
return aperture_mask_method
@staticmethod
def _validate_kron_params(kron_params):
if np.ndim(kron_params) != 1:
msg = 'kron_params must be 1D'
raise ValueError(msg)
nparams = len(kron_params)
if nparams not in (2, 3):
msg = 'kron_params must have 2 or 3 elements'
raise ValueError(msg)
if kron_params[0] <= 0:
msg = 'kron_params[0] must be > 0'
raise ValueError(msg)
if kron_params[1] <= 0:
msg = 'kron_params[1] must be > 0'
raise ValueError(msg)
if nparams == 3 and kron_params[2] < 0:
msg = 'kron_params[2] must be >= 0'
raise ValueError(msg)
return tuple(kron_params)
def _validate_detection_catalog(self, detection_catalog):
if detection_catalog is None:
return None
if not isinstance(detection_catalog, SourceCatalog):
msg = 'detection_catalog must be a SourceCatalog instance'
raise TypeError(msg)
if not np.array_equal(detection_catalog._segmentation_image,
self._segmentation_image):
msg = ('detection_catalog must have same segmentation_image '
'as the input segmentation_image')
raise ValueError(msg)
return detection_catalog
def _update_meta(self):
meta_values = {}
attrs = ('local_bkg_width', 'aperture_mask_method', 'kron_params')
for attr in attrs:
meta_values[attr] = getattr(self, attr)
if not _get_future_column_names():
for old_name, new_name in _DEPRECATED_META_KEYS.items():
if new_name in meta_values:
meta_values[old_name] = meta_values[new_name]
self.meta.update(meta_values)
def _set_semode(self):
# SE emulation
self._aperture_mask_kwargs = {
'circ': {'method': 'subpixel', 'subpixels': 5},
'kron': {'method': 'center'},
'flux_radius': {'method': 'subpixel', 'subpixels': 5},
'cen_win': {'method': 'subpixel', 'subpixels': 11},
}
@property
def _properties(self):
"""
A list of all class properties, include lazyproperties (even in
superclasses).
The result is cached on the class to avoid repeated
introspection via `inspect.getmembers`.
"""
cls = self.__class__
attr = '_cached_properties'
# Subclasses get their own property list
if attr not in cls.__dict__:
def isproperty(obj):
return isinstance(obj, property)
setattr(cls, attr,
[i[0] for i in inspect.getmembers(
cls, predicate=isproperty)])
return getattr(cls, attr)
@property
def properties(self):
"""
A list of built-in source properties.
"""
lazyproperties = [name for name in self._lazyproperties if not
name.startswith('_')]
lazyproperties.remove('isscalar')
lazyproperties.remove('n_labels')
lazyproperties.extend(['label', 'labels', 'slices'])
lazyproperties.sort()
return lazyproperties
@property
def _lazyproperties(self):
"""
A list of all class lazyproperties (even in superclasses).
The result is cached on the class to avoid repeated
introspection via `inspect.getmembers`.
"""
cls = self.__class__
attr = '_cached_lazyproperties'
# Subclasses get their own lazyproperty list
if attr not in cls.__dict__:
def islazyproperty(obj):
return isinstance(obj, lazyproperty)
setattr(cls, attr,
[i[0] for i in inspect.getmembers(
cls, predicate=islazyproperty)])
return getattr(cls, attr)
@staticmethod
def _index_object_list(lst, index):
"""
Index a list of heterogeneous objects using numpy-style
indexing.
A numpy object array is used to support fancy and boolean
indices on lists of tuples or other structured objects.
A sentinel ``None`` is appended (and then removed) to prevent
numpy from recursing into nested sequences (e.g., tuples of
slices).
Parameters
----------
lst : list
The list of objects to index.
index : int, slice, list, or array
The index to apply to the list.
Returns
-------
result : list or object
A list for array results or the element itself for scalar
(integer) indices.
"""
result = np.array([*lst, None], dtype=object)[:-1][index]
if isinstance(result, np.ndarray):
return result.tolist()
return result
def __getitem__(self, index):
if self.isscalar:
msg = (f'A scalar {self.__class__.__name__!r} object cannot '
'be indexed')
raise TypeError(msg)
newcls = object.__new__(self.__class__)
# Attributes defined in __init__ that are copied directly to the
# new class
init_attr = ('_data', '_segmentation_image', '_error', '_mask',
'_background', 'wcs', '_data_unit', '_convolved_data',
'local_bkg_width', 'aperture_mask_method',
'kron_params', 'default_columns', '_custom_properties',
'meta', '_aperture_mask_kwargs', 'progress_bar')
for attr in init_attr:
setattr(newcls, attr, getattr(self, attr))
# _labels determines ordering and isscalar
attr = '_labels'
setattr(newcls, attr, getattr(self, attr)[index])
# Need to slice detection_catalog, if input
attr = '_detection_catalog'
if getattr(self, attr) is None:
setattr(newcls, attr, None)
else:
setattr(newcls, attr, getattr(self, attr)[index])
attr = '_slices'
value = self._index_object_list(getattr(self, attr), index)
setattr(newcls, attr, value)
# Slice the flux_radius cache values
newcls._flux_radius_cache = {key: value[index]
for key, value
in self._flux_radius_cache.items()}
# Evaluated lazyproperty objects and extra properties
keys = (set(self.__dict__.keys())
& (set(self._lazyproperties) | set(self._custom_properties)))
for key in keys:
value = self.__dict__[key]
# Do not insert attributes that are always scalar (e.g.,
# isscalar, n_labels), i.e., not an array/list for each
# source
if np.isscalar(value):
continue
try:
# Keep _<attr> lazyproperties as length-1 iterables;
# _<attr> lazyproperties should not have @as_scalar applied
if newcls.isscalar and key.startswith('_'):
if isinstance(value, np.ndarray):
val = value[:, np.newaxis][index]
else:
val = [value[index]]
else:
val = value[index]
except TypeError:
# Apply fancy indices (e.g., array/list or bool
# mask) to lists
val = self._index_object_list(value, index)
newcls.__dict__[key] = val
return newcls
def __str__(self):
cls_name = f'<{self.__class__.__module__}.{self.__class__.__name__}>'
with np.printoptions(threshold=25, edgeitems=5):
fmt = [f'Length: {self.n_labels}', f'labels: {self.labels}']
return f'{cls_name}\n' + '\n'.join(fmt)
def __repr__(self):
return self.__str__()
def __len__(self):
if self.isscalar:
msg = f'Scalar {self.__class__.__name__!r} object has no len()'
raise TypeError(msg)
return self.n_labels
def __iter__(self):
for item in range(len(self)):
yield self.__getitem__(item)
# Remove in 4.0
def __getattr__(self, name):
return deprecated_getattr(self, name, _DEPRECATED_ATTRIBUTES,
since='3.0', until='4.0')
@lazyproperty
def isscalar(self):
"""
Whether the instance is scalar (e.g., a single source).
"""
return self._labels.shape == ()
@staticmethod
def _has_len(value):
if isinstance(value, str):
return False
try:
# NOTE: cannot just check for __len__ attribute, because
# it could exist, but raise an Exception for scalar objects
len(value)
except TypeError:
return False
return True
[docs]
def copy(self):
"""
Return a deep copy of this SourceCatalog.
Returns
-------
result : `SourceCatalog`
A deep copy of this object.
"""
return deepcopy(self)
@property
def custom_properties(self):
"""
A list of the user-defined source properties.
"""
return self._custom_properties
[docs]
@deprecated_positional_kwargs(since='3.0', until='4.0')
def add_property(self, name, value, overwrite=False):
"""
Add a user-defined property as a new attribute.
For example, the property ``name`` can then be included in the
`to_table` ``columns`` keyword list to output the results in the
table.
The complete list of user-defined properties is stored in the
`custom_properties` attribute.
Parameters
----------
name : str
The name of property. The name must not conflict with any of
the built-in property names or attributes.
value : array_like or float
The value to assign.
overwrite : bool, option
If `True`, will overwrite the existing property ``name``.
"""
internal_attributes = ((set(self.__dict__.keys())
| set(self._properties))
- set(self.custom_properties))
if name in internal_attributes:
msg = f'{name} cannot be set because it is a built-in attribute'
raise ValueError(msg)
if not overwrite:
if hasattr(self, name):
msg = (f'{name} already exists as an attribute. Set '
'overwrite=True to overwrite an existing attribute.')
raise ValueError(msg)
if name in self._custom_properties:
msg = (f'{name} already exists in the custom_properties '
'attribute list.')
raise ValueError(msg)
property_error = False
if self.isscalar:
# This allows flux_radius to add len-1 array values for
# scalar self
if self._has_len(value) and len(value) == 1:
value = value[0]
if hasattr(value, 'isscalar'):
# e.g., Quantity, SkyCoord, Time
if not value.isscalar:
property_error = True
elif not np.isscalar(value):
property_error = True
elif not self._has_len(value) or len(value) != self.n_labels:
property_error = True
if property_error:
msg = ('value must have the same number of elements as the '
'catalog in order to add it as a property.')
raise ValueError(msg)
setattr(self, name, value)
if name not in self._custom_properties:
self._custom_properties.append(name)
[docs]
def remove_property(self, name):
"""
Remove a user-defined property.
The property must have been defined using `add_property`. The
complete list of user-defined properties is stored in the
`custom_properties` attribute.
Parameters
----------
name : str
The name of the property to remove.
"""
self.remove_properties(name)
[docs]
def remove_properties(self, names):
"""
Remove user-defined properties.
The properties must have been defined using `add_property`.
The complete list of user-defined properties is stored in the
`custom_properties` attribute.
Parameters
----------
names : list of str or str
The names of the properties to remove.
"""
if isinstance(names, str):
names = [names]
# We copy the list here to prevent changing the list in-place
# during the for loop below, e.g., in case a user inputs
# self.custom_properties to ``names``
custom_properties = self._custom_properties.copy()
for name in names:
if name in custom_properties:
delattr(self, name)
custom_properties.remove(name)
else:
msg = f'{name} is not a defined property'
raise ValueError(msg)
self._custom_properties = custom_properties
[docs]
def rename_property(self, name, new_name):
"""
Rename a user-defined property.
The renamed property will remain at the same index in the
`custom_properties` list.
Parameters
----------
name : str
The old attribute name.
new_name : str
The new attribute name.
"""
self.add_property(new_name, getattr(self, name))
idx = self.custom_properties.index(name)
self.remove_property(name)
# Preserve the order of self.custom_properties
self.custom_properties.remove(new_name)
self.custom_properties.insert(idx, new_name)
@lazyproperty
def _null_objects(self):
"""
Return `None` values.
For example, this is used for SkyCoord properties if ``wcs`` is
`None`.
"""
return np.array([None] * self.n_labels)
@lazyproperty
def _null_values(self):
"""
Return np.nan values.
For example, this is used for background properties if
``background`` is `None`.
"""
values = np.empty(self.n_labels)
values.fill(np.nan)
return values
@lazyproperty
def _data_cutouts(self):
"""
A list of data cutouts using the segmentation image slices.
"""
return [self._data[slc] for slc in self._slices_iter]
@lazyproperty
def _segmentation_image_cutouts(self):
"""
A list of segmentation image cutouts using the segmentation
image slices.
"""
return [self._segmentation_image.data[slc]
for slc in self._slices_iter]
@lazyproperty
def _mask_cutouts(self):
"""
A list of mask cutouts using the segmentation image slices.
If the input ``mask`` is None then a list of None is returned.
"""
if self._mask is None:
return self._null_objects
return [self._mask[slc] for slc in self._slices_iter]
@lazyproperty
def _error_cutouts(self):
"""
A list of error cutouts using the segmentation image slices.
If the input ``error`` is None then a list of None is returned.
"""
if self._error is None:
return self._null_objects
return [self._error[slc] for slc in self._slices_iter]
@lazyproperty
def _convdata_cutouts(self):
"""
A list of convolved data cutouts using the segmentation image
slices.
"""
return [self._convolved_data[slc] for slc in self._slices_iter]
@lazyproperty
def _background_cutouts(self):
"""
A list of background cutouts using the segmentation image
slices.
"""
if self._background is None:
return self._null_objects
return [self._background[slc] for slc in self._slices_iter]
@staticmethod
def _make_cutout_data_mask(data_cutout, mask_cutout):
"""
Make a cutout data mask, combining both the input ``mask`` and
non-finite ``data`` values.
"""
data_mask = ~np.isfinite(data_cutout)
if mask_cutout is not None:
data_mask |= mask_cutout
return data_mask
def _make_cutout_data_masks(self, data_cutouts, mask_cutouts):
"""
Make a list of cutout data masks, combining both the input
``mask`` and non-finite ``data`` values for each source.
"""
data_masks = []
for (data_cutout, mask_cutout) in zip(data_cutouts, mask_cutouts,
strict=True):
data_masks.append(self._make_cutout_data_mask(data_cutout,
mask_cutout))
return data_masks
@lazyproperty
def _cutout_segment_masks(self):
"""
Cutout boolean mask for source segment.
The mask is `True` for all pixels (background and from other
source segments) outside the source segment.
"""
return [segm != label
for label, segm
in zip(self.labels,
self._segmentation_image_cutouts,
strict=True)]
@lazyproperty
def _cutout_data_masks(self):
"""
Cutout boolean mask of non-finite ``data`` values combined with
the input ``mask`` array.
The mask is `True` for non-finite ``data`` values and where the
input ``mask`` is `True`.
"""
return self._make_cutout_data_masks(self._data_cutouts,
self._mask_cutouts)
@lazyproperty
def _cutout_total_masks(self):
"""
Boolean mask representing the combination of
``_cutout_segment_masks`` and ``_cutout_data_masks``.
This mask is applied to ``data``, ``error``, and ``background``
inputs when calculating properties.
"""
masks = []
for mask1, mask2 in zip(self._cutout_segment_masks,
self._cutout_data_masks, strict=True):
masks.append(mask1 | mask2)
return masks
@lazyproperty
def _moment_data_cutouts(self):
"""
A list of 2D `~numpy.ndarray` cutouts from the (convolved) data.
The following pixels are set to zero in these arrays:
* pixels outside the source segment
* any masked pixels from the input ``mask``
* invalid convolved data values (NaN and inf)
* negative convolved data values; negative pixels (especially
at large radii) can give image moments that have negative
variances.
These arrays are used to derive moment-based properties.
"""
cutouts = []
for convdata_cutout, mask_cutout, segmmask_cutout in zip(
self._convdata_cutouts, self._mask_cutouts,
self._cutout_segment_masks, strict=True):
convdata_mask = (~np.isfinite(convdata_cutout)
| (convdata_cutout < 0) | segmmask_cutout)
if self._mask is not None:
convdata_mask |= mask_cutout
cutout = convdata_cutout.copy()
cutout[convdata_mask] = 0.0
cutouts.append(cutout)
return cutouts
def _prepare_cutouts(self, arrays, *, units=True, masked=False,
dtype=None):
"""
Prepare cutouts by applying optional units, masks, or dtype.
"""
if units and masked:
msg = 'Both units and masked cannot be True'
raise ValueError(msg)
if dtype is not None:
cutouts = [cutout.astype(dtype, copy=True) for cutout in arrays]
else:
cutouts = arrays
if units and self._data_unit is not None:
cutouts = [(cutout << self._data_unit) for cutout in cutouts]
if masked:
return [np.ma.masked_array(cutout, mask=mask)
for cutout, mask in zip(cutouts, self._cutout_total_masks,
strict=True)]
return cutouts
[docs]
def select_label(self, label):
"""
Return a new `SourceCatalog` object for the input ``label``
only.
Parameters
----------
label : int
The source label.
Returns
-------
cat : `SourceCatalog`
A new `SourceCatalog` object containing only the source with
the input ``label``.
"""
return self.select_labels(label)
[docs]
def select_labels(self, labels):
"""
Return a new `SourceCatalog` object for the input ``labels``
only.
Parameters
----------
labels : list, tuple, or `~numpy.ndarray` of int
The source label(s).
Returns
-------
cat : `SourceCatalog`
A new `SourceCatalog` object containing only the sources with
the input ``labels``.
"""
self._segmentation_image.check_labels(labels)
sorter = np.argsort(self.labels)
indices = sorter[np.searchsorted(self.labels, labels, sorter=sorter)]
return self[indices]
[docs]
@deprecated_positional_kwargs(since='3.0', until='4.0')
def to_table(self, columns=None):
"""
Create a `~astropy.table.QTable` of source properties.
Parameters
----------
columns : str, list of str, `None`, optional
Names of columns, in order, to include in the output
`~astropy.table.QTable`. The allowed column names are any of
the `SourceCatalog` properties or custom properties added
using `add_property`. If ``columns`` is `None`, then a
default list of scalar-valued properties (as defined by the
``default_columns`` attribute) will be used.
Returns
-------
table : `~astropy.table.QTable`
A table of sources properties with one row per source.
"""
if columns is None:
table_columns = self.default_columns
elif isinstance(columns, str):
table_columns = [columns]
else:
table_columns = columns
# Replace with QTable() in 4.0
tbl = create_empty_deprecated_qtable(
_DEPRECATED_ATTRIBUTES, since='3.0', until='4.0')
tbl.meta.update(self.meta) # keep tbl.meta type
for column in table_columns:
values = getattr(self, column)
# Column assignment requires an object with a length
if self.isscalar:
values = (values,)
tbl[column] = values
return tbl
@lazyproperty
def n_labels(self):
"""
The number of source labels.
"""
return len(self.labels)
@property
@as_scalar
def label(self):
"""
The source label number(s).
This label number corresponds to the assigned pixel value in the
`~photutils.segmentation.SegmentationImage`.
Returns an array for multi-source catalogs, or a scalar for a
single-source catalog.
"""
return self._labels
@property
def labels(self):
"""
The source label number(s), always as an iterable
`~numpy.ndarray`.
This label number corresponds to the assigned pixel value in the
`~photutils.segmentation.SegmentationImage`.
"""
labels = self.label
if self.isscalar:
labels = np.array((labels,))
return labels
@property
@as_scalar
def slices(self):
"""
A tuple of slice objects defining the minimal bounding box of
the source.
Returns a list for multi-source catalogs, or a single tuple for
a single-source catalog.
"""
return self._slices
@lazyproperty
def _slices_iter(self):
"""
A tuple of slice objects defining the minimal bounding box of
the source, always as an iterable.
"""
_slices = self.slices
if self.isscalar:
_slices = (_slices,)
return _slices
@lazyproperty
@as_scalar
def segment_cutout(self):
"""
A 2D `~numpy.ndarray` cutout of the segmentation image using the
minimal bounding box of the source.
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
return self._prepare_cutouts(
self._segmentation_image_cutouts, units=False,
masked=False)
@lazyproperty
@as_scalar
def segment_cutout_masked(self):
"""
A 2D `~numpy.ma.MaskedArray` cutout of the segmentation image
using the minimal bounding box of the source.
The mask is `True` for pixels outside the source segment
(labeled region of interest), masked pixels from the ``mask``
input, or any non-finite ``data`` values (NaN and inf).
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
return self._prepare_cutouts(
self._segmentation_image_cutouts, units=False,
masked=True)
@lazyproperty
@as_scalar
def data_cutout(self):
"""
A 2D `~numpy.ndarray` cutout from the data using the minimal
bounding box of the source.
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
return self._prepare_cutouts(self._data_cutouts, units=True,
masked=False, dtype=float)
@lazyproperty
@as_scalar
def data_cutout_masked(self):
"""
A 2D `~numpy.ma.MaskedArray` cutout from the data using the
minimal bounding box of the source.
The mask is `True` for pixels outside the source segment
(labeled region of interest), masked pixels from the ``mask``
input, or any non-finite ``data`` values (NaN and inf).
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
return self._prepare_cutouts(self._data_cutouts, units=False,
masked=True, dtype=float)
@lazyproperty
@as_scalar
def conv_data_cutout(self):
"""
A 2D `~numpy.ndarray` cutout from the convolved data using the
minimal bounding box of the source.
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
return self._prepare_cutouts(self._convdata_cutouts, units=True,
masked=False, dtype=float)
@lazyproperty
@as_scalar
def conv_data_cutout_masked(self):
"""
A 2D `~numpy.ma.MaskedArray` cutout from the convolved data
using the minimal bounding box of the source.
The mask is `True` for pixels outside the source segment
(labeled region of interest), masked pixels from the ``mask``
input, or any non-finite ``data`` values (NaN and inf).
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
return self._prepare_cutouts(self._convdata_cutouts, units=False,
masked=True, dtype=float)
@lazyproperty
@as_scalar
def error_cutout(self):
"""
A 2D `~numpy.ndarray` cutout from the error array using the
minimal bounding box of the source.
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
if self._error is None:
return self._null_objects
return self._prepare_cutouts(self._error_cutouts, units=True,
masked=False)
@lazyproperty
@as_scalar
def error_cutout_masked(self):
"""
A 2D `~numpy.ma.MaskedArray` cutout from the error array using
the minimal bounding box of the source.
The mask is `True` for pixels outside the source segment
(labeled region of interest), masked pixels from the ``mask``
input, or any non-finite ``data`` values (NaN and inf).
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
if self._error is None:
return self._null_objects
return self._prepare_cutouts(self._error_cutouts, units=False,
masked=True)
@lazyproperty
@as_scalar
def background_cutout(self):
"""
A 2D `~numpy.ndarray` cutout from the background array using the
minimal bounding box of the source.
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
if self._background is None:
return self._null_objects
return self._prepare_cutouts(self._background_cutouts, units=True,
masked=False)
@lazyproperty
@as_scalar
def background_cutout_masked(self):
"""
A 2D `~numpy.ma.MaskedArray` cutout from the background array.
using the minimal bounding box of the source.
The mask is `True` for pixels outside the source segment
(labeled region of interest), masked pixels from the ``mask``
input, or any non-finite ``data`` values (NaN and inf).
Returns a list of arrays for multi-source catalogs, or a single
array for a single-source catalog.
"""
if self._background is None:
return self._null_objects
return self._prepare_cutouts(self._background_cutouts, units=False,
masked=True)
@lazyproperty
def _all_masked(self):
"""
True if all pixels over the source segment are masked.
"""
return np.array([np.all(mask) for mask in self._cutout_total_masks])
def _get_values(self, array):
"""
Get a 1D array of unmasked values from the input array within
the source segment.
An array with a single NaN is returned for completely-masked
sources.
"""
if self.isscalar:
array = (array,)
return [arr.compressed() if len(arr.compressed()) > 0
else np.array([np.nan]) for arr in array]
@staticmethod
def _reduceat(values, ufunc, *, transform=None):
"""
Apply ``ufunc.reduceat`` to a list of arrays.
This is significantly faster than a list comprehension with
individual NumPy calls for each array.
Parameters
----------
values : list of 1D `~numpy.ndarray`
A list of 1D arrays.
ufunc : `~numpy.ufunc`
The NumPy ufunc to apply (e.g., `~numpy.add`,
`~numpy.minimum`, `~numpy.maximum`).
transform : callable or None, optional
An optional transformation to apply to the concatenated
array before reducing (e.g., `~numpy.square`).
Returns
-------
result : `~numpy.ndarray`
The reduceat result.
sizes : `~numpy.ndarray`
The sizes of the input arrays.
"""
if not values:
return np.array([]), np.array([], dtype=int)
sizes = np.array([len(arr) for arr in values])
splits = np.concatenate(([0], np.cumsum(sizes[:-1])))
concat = np.concatenate(values)
if transform is not None:
concat = transform(concat)
return ufunc.reduceat(concat, splits), sizes
@lazyproperty
def _data_values(self):
"""
A 1D array of unmasked data values.
An array with a single NaN is returned for completely-masked
sources.
"""
return self._get_values(self.data_cutout_masked)
@lazyproperty
def _error_values(self):
"""
A 1D array of unmasked error values.
An array with a single NaN is returned for completely-masked
sources.
"""
if self._error is None:
return self._null_objects
return self._get_values(self.error_cutout_masked)
@lazyproperty
def _background_values(self):
"""
A 1D array of unmasked background values.
An array with a single NaN is returned for completely-masked
sources.
"""
if self._background is None:
return self._null_objects
return self._get_values(self.background_cutout_masked)
@lazyproperty
@use_detcat
@as_scalar
def moments(self):
"""
Spatial moments up to 3rd order of the source.
"""
result = []
for arr in self._moment_data_cutouts:
ny, nx = arr.shape
y = np.arange(ny, dtype=float)
x = np.arange(nx, dtype=float)
yp = np.column_stack([np.ones(ny), y, y * y, y ** 3])
xp = np.column_stack([np.ones(nx), x, x * x, x ** 3])
result.append(yp.T @ arr @ xp)
return np.array(result)
@lazyproperty
@use_detcat
@as_scalar
def moments_central(self):
"""
Central moments (translation invariant) of the source up to 3rd
order.
"""
cutout_centroid = self.cutout_centroid
if self.isscalar:
cutout_centroid = cutout_centroid[np.newaxis, :]
result = []
for arr, xcen, ycen in zip(self._moment_data_cutouts,
cutout_centroid[:, 0],
cutout_centroid[:, 1], strict=True):
ny, nx = arr.shape
yc = np.arange(ny, dtype=float) - ycen
xc = np.arange(nx, dtype=float) - xcen
yp = np.column_stack([np.ones(ny), yc, yc * yc, yc ** 3])
xp = np.column_stack([np.ones(nx), xc, xc * xc, xc ** 3])
result.append(yp.T @ arr @ xp)
return np.array(result)
@lazyproperty
@use_detcat
@as_scalar
def cutout_centroid(self):
"""
The ``(x, y)`` coordinate, relative to the cutout data, of the
centroid within the isophotal source segment.
The centroid is computed as the center of mass of the unmasked
pixels within the source segment.
"""
moments = self.moments
if self.isscalar:
moments = moments[np.newaxis, :]
# Ignore divide-by-zero RuntimeWarning
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
y_centroid = moments[:, 1, 0] / moments[:, 0, 0]
x_centroid = moments[:, 0, 1] / moments[:, 0, 0]
return np.transpose((x_centroid, y_centroid))
@lazyproperty
@use_detcat
@as_scalar
def centroid(self):
"""
The ``(x, y)`` coordinate of the centroid within the isophotal
source segment.
The centroid is computed as the center of mass of the unmasked
pixels within the source segment.
"""
origin = np.transpose((self.bbox_xmin, self.bbox_ymin))
return self.cutout_centroid + origin
@lazyproperty
@use_detcat
def _x_centroid(self):
"""
The ``x`` coordinate of the `centroid` within the source
segment, always as an iterable.
"""
if self.isscalar:
x_centroid = self.centroid[0:1] # scalar array
else:
x_centroid = self.centroid[:, 0]
return x_centroid
@lazyproperty
@use_detcat
@as_scalar
def x_centroid(self):
"""
The ``x`` coordinate of the `centroid` within the source
segment.
The centroid is computed as the center of mass of the unmasked
pixels within the source segment.
"""
return self._x_centroid
@lazyproperty
@use_detcat
def _y_centroid(self):
"""
The ``y`` coordinate of the `centroid` within the source
segment, always as an iterable.
"""
if self.isscalar:
y_centroid = self.centroid[1:2] # scalar array
else:
y_centroid = self.centroid[:, 1]
return y_centroid
@lazyproperty
@use_detcat
@as_scalar
def y_centroid(self):
"""
The ``y`` coordinate of the `centroid` within the source
segment.
The centroid is computed as the center of mass of the unmasked
pixels within the source segment.
"""
return self._y_centroid
@lazyproperty
@use_detcat
@as_scalar
def centroid_win(self):
"""
The ``(x, y)`` coordinate of the "windowed" centroid.
The window centroid is computed using an iterative algorithm
to derive a more accurate centroid. It is equivalent to
`SourceExtractor`_'s XWIN_IMAGE and YWIN_IMAGE parameters.
Notes
-----
During each iteration, the centroid is calculated using all
pixels within a circular aperture of ``4 * sigma`` from the
current position, weighting pixel values with a 2D Gaussian with
a standard deviation of ``sigma``. ``sigma`` is the half-light
radius (i.e., ``flux_radius(0.5)``) times (2.0 / 2.35). A
minimum half-light radius of 0.5 pixels is used. Iteration stops
when the change in centroid position falls below a pre-defined
threshold or a maximum number of iterations is reached.
If the windowed centroid falls outside the 1-sigma ellipse
shape based on the image moments, then the isophotal `centroid`
will be used instead. If the half-light radius is not finite
(e.g., due to a non-finite Kron radius), then ``np.nan`` will be
returned.
"""
# Use .copy() to avoid mutating the cached flux_radius value
radius_hl = self.flux_radius(0.5).value.copy()
if self.isscalar:
radius_hl = np.array([radius_hl])
# Track which sources have non-finite half-light radii (e.g.,
# due to NaN kron_radius). These sources cannot have a
# meaningful windowed centroid.
nan_hl = ~np.isfinite(radius_hl)
# Apply a minimum half-light radius of 0.5 pixels (matching
# SourceExtractor) for valid but very small values
min_radius = 0.5
small_mask = np.isfinite(radius_hl) & (radius_hl < min_radius)
radius_hl[small_mask] = min_radius
labels = self.labels
if self.progress_bar:
desc = 'centroid_win'
labels = add_progress_bar(labels, desc=desc)
# Pre-fetch arrays used in the inner loop
data_arr = self._data
mask_arr = self._mask
segm_data = self._segmentation_image.data
data_shape = data_arr.shape
do_correct = self.aperture_mask_method == 'correct'
do_segm_mask = self.aperture_mask_method != 'none'
max_aper_size = max(data_arr.size, 1_000_000)
max_iters = 16
centroid_threshold = 0.0001
xcen_win = []
ycen_win = []
for label, xcen, ycen, rad_hl, nan_hl_ in zip(
labels, self._x_centroid, self._y_centroid, radius_hl,
nan_hl, strict=True):
if nan_hl_ or math.isnan(xcen) or math.isnan(ycen):
xcen_win.append(np.nan)
ycen_win.append(np.nan)
continue
sigma = 2.0 * rad_hl * gaussian_fwhm_to_sigma
inv_2sigma2 = -1.0 / (2.0 * sigma * sigma)
radius = 4.0 * sigma
radius_sq = radius * radius
# Compute the full (unclipped) bounding box for the aperture
# using the initial centroid. The radius is fixed, so the
# bbox size stays the same across iterations even if the
# center shifts slightly.
bbox_halfsize = int(radius + 1.5)
full_ny = full_nx = 2 * bbox_halfsize + 1
# OOM guard
if full_ny * full_nx > max_aper_size:
xcen_win.append(np.nan)
ycen_win.append(np.nan)
continue
# Cache for cutout data when the integer bbox doesn't change
prev_ixcen = prev_iycen = None
cached_data = None
cached_mask = None
iter_ = 0
dcen = 1.0
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
while iter_ < max_iters and dcen > centroid_threshold:
# Compute integer bounding box
ixmin = int(xcen + 0.5) - bbox_halfsize
ixmax = ixmin + full_nx
iymin = int(ycen + 0.5) - bbox_halfsize
iymax = iymin + full_ny
# Clip to data boundaries
slc_y = slice(max(0, iymin), min(data_shape[0], iymax))
slc_x = slice(max(0, ixmin), min(data_shape[1], ixmax))
if (slc_y.start >= slc_y.stop
or slc_x.start >= slc_x.stop):
xcen = np.nan
ycen = np.nan
break
cur_ixcen = int(xcen + 0.5)
cur_iycen = int(ycen + 0.5)
# Recompute cutout data only when the integer center
# changes to avoid redundant _mask_to_mirrored_value
# calls
if cur_ixcen != prev_ixcen or cur_iycen != prev_iycen:
prev_ixcen = cur_ixcen
prev_iycen = cur_iycen
data = data_arr[slc_y, slc_x].astype(float)
data_mask = ~np.isfinite(data)
if mask_arr is not None:
data_mask |= mask_arr[slc_y, slc_x]
cutout_xycen = (xcen - max(0, ixmin),
ycen - max(0, iymin))
if do_segm_mask:
seg_cut = segm_data[slc_y, slc_x]
segm_mask = ((seg_cut != label)
& (seg_cut != 0))
if self.aperture_mask_method == 'mask':
data_mask = data_mask | segm_mask
if do_correct:
data = _mask_to_mirrored_value(
data, segm_mask, cutout_xycen,
mask=data_mask)
cached_data = data
cached_mask = data_mask
# Centroid position in cutout coordinates
cx = xcen - max(0, ixmin)
cy = ycen - max(0, iymin)
ny = slc_y.stop - slc_y.start
nx = slc_x.stop - slc_x.start
# Build coordinate grids relative to centroid
# (reused for circle mask, Gaussian, and moments)
xvals = np.arange(nx) - cx
yvals = np.arange(ny) - cy
xx = xvals[np.newaxis, :]
yy = yvals[:, np.newaxis]
# Inline binary circle mask
rr2 = xx * xx + yy * yy
aper_weights = (rr2 <= radius_sq).astype(float)
# Inline Gaussian weight
gweight = np.exp(rr2 * inv_2sigma2)
# Apply weights and mask
weighted = (cached_data * aper_weights * gweight)
weighted[cached_mask] = 0.0
# Inline moment computation
total = np.sum(weighted)
dx = np.sum(weighted * xx) / total
dy = np.sum(weighted * yy) / total
dcen = math.sqrt(dx * dx + dy * dy)
xcen += dx * 2.0
ycen += dy * 2.0
iter_ += 1
xcen_win.append(xcen)
ycen_win.append(ycen)
xcen_win = np.array(xcen_win)
ycen_win = np.array(ycen_win)
# Reset to the isophotal centroid if the windowed centroid is
# outside the 1-sigma ellipse or if the iteration failed (NaN
# from aperture off-image). Sources with NaN half-light radius
# keep NaN (no valid window size).
dx = self._x_centroid - xcen_win
dy = self._y_centroid - ycen_win
cxx = self.ellipse_cxx.value
cxy = self.ellipse_cxy.value
cyy = self.ellipse_cyy.value
if self.isscalar:
cxx = (cxx,)
cxy = (cxy,)
cyy = (cyy,)
reset = ((cxx * dx**2 + cxy * dx * dy + cyy * dy**2) > 1)
nan_cen = np.isnan(xcen_win) | np.isnan(ycen_win)
reset |= nan_cen & ~nan_hl
if np.any(reset):
xcen_win[reset] = self._x_centroid[reset]
ycen_win[reset] = self._y_centroid[reset]
return np.transpose((xcen_win, ycen_win))
@lazyproperty
@use_detcat
@as_scalar
def x_centroid_win(self):
"""
The ``x`` coordinate of the "windowed" centroid
(`centroid_win`).
The window centroid is computed using an iterative algorithm
to derive a more accurate centroid. It is equivalent to
`SourceExtractor`_'s XWIN_IMAGE parameters.
"""
if self.isscalar:
x_centroid = self.centroid_win[0] # scalar array
else:
x_centroid = self.centroid_win[:, 0]
return x_centroid
@lazyproperty
@use_detcat
@as_scalar
def y_centroid_win(self):
"""
The ``y`` coordinate of the "windowed" centroid
(`centroid_win`).
The window centroid is computed using an iterative algorithm
to derive a more accurate centroid. It is equivalent to
`SourceExtractor`_'s YWIN_IMAGE parameters.
"""
if self.isscalar:
y_centroid = self.centroid_win[1] # scalar array
else:
y_centroid = self.centroid_win[:, 1]
return y_centroid
@lazyproperty
@use_detcat
@as_scalar
def cutout_centroid_win(self):
"""
The ``(x, y)`` coordinate, relative to the cutout data, of the
"windowed" centroid.
The window centroid is computed using an iterative algorithm
to derive a more accurate centroid. It is equivalent to
`SourceExtractor`_'s XWIN_IMAGE and YWIN_IMAGE parameters. See
`centroid_win` for further details about the algorithm.
"""
origin = np.transpose((self.bbox_xmin, self.bbox_ymin))
return self.centroid_win - origin
@lazyproperty
@use_detcat
@as_scalar
def cutout_centroid_quad(self):
"""
The ``(x, y)`` centroid coordinate, relative to the cutout data,
calculated by fitting a 2D quadratic polynomial to the unmasked
pixels in the source segment.
Notes
-----
`~photutils.centroids.centroid_quadratic` is used to calculate
the centroid with ``fit_boxsize=3``.
Because this centroid is based on fitting data, it can fail for
many reasons including:
* quadratic fit failed
* quadratic fit does not have a maximum
* quadratic fit maximum falls outside image
* not enough unmasked data points (6 are required)
In these cases, then the isophotal `centroid` will be used
instead.
Also note that a fit is not performed if the maximum data value
is at the edge of the source segment. In this case, the position
of the maximum pixel will be returned.
"""
# Precompute the pseudo-inverse for the 3x3 relative coordinate
# design matrix [1, x, y, xy, x^2, y^2]. This is constant for
# all sources and avoids per-source lstsq calls.
xi = np.arange(3)
x, y = np.meshgrid(xi, xi)
x = x.ravel()
y = y.ravel()
coeff_matrix = np.empty((9, 6), dtype=float)
coeff_matrix[:, 0] = 1
coeff_matrix[:, 1] = x
coeff_matrix[:, 2] = y
coeff_matrix[:, 3] = x * y
coeff_matrix[:, 4] = x * x
coeff_matrix[:, 5] = y * y
pinv = np.linalg.pinv(coeff_matrix)
_nan = np.nan
centroid_quad = []
cutouts = self._data_cutouts
if self.progress_bar:
desc = 'centroid_quad'
cutouts = add_progress_bar(cutouts, desc=desc)
for cutout, mask in zip(cutouts, self._cutout_total_masks,
strict=True):
ny, nx = cutout.shape
# Cutout must be at least 3x3 for the quadratic fit
if ny < 3 or nx < 3:
centroid_quad.append((_nan, _nan))
continue
# Apply mask: _cutout_total_masks already includes
# non-finite data values, so cutout[mask] = 0.0 handles both
# masked pixels and non-finite values.
cutout = np.array(cutout, dtype=float)
cutout[mask] = 0.0
# Find peak pixel
yidx, xidx = np.unravel_index(np.argmax(cutout), cutout.shape)
# If peak at edge of cutout, return peak position
if xidx == 0 or xidx == nx - 1 or yidx == 0 or yidx == ny - 1:
centroid_quad.append((float(xidx), float(yidx)))
continue
# Extract 3x3 box centered on peak (guaranteed to fit
# since peak is not at edge)
xidx0 = xidx - 1
yidx0 = yidx - 1
cutout_flat = cutout[yidx0:yidx0 + 3, xidx0:xidx0 + 3].ravel()
# Compute polynomial coefficients via precomputed
# pseudo-inverse
c = pinv @ cutout_flat
c10, c01, c11, c20, c02 = c[1], c[2], c[3], c[4], c[5]
det = 4.0 * c20 * c02 - c11 * c11
if det <= 0 or c20 > 0:
centroid_quad.append((_nan, _nan))
continue
# Maximum in relative coords, then convert to cutout coords
xm = (c01 * c11 - 2.0 * c02 * c10) / det + xidx0
ym = (c10 * c11 - 2.0 * c20 * c01) / det + yidx0
if 0.0 < xm < (nx - 1.0) and 0.0 < ym < (ny - 1.0):
centroid_quad.append((xm, ym))
else:
centroid_quad.append((_nan, _nan))
centroid_quad = np.array(centroid_quad)
# Use the segment barycenter if fit returned NaN
nan_mask = (np.isnan(centroid_quad[:, 0])
| np.isnan(centroid_quad[:, 1]))
if np.any(nan_mask):
centroid_quad[nan_mask] = self.cutout_centroid[nan_mask]
return centroid_quad
@lazyproperty
@use_detcat
@as_scalar
def centroid_quad(self):
"""
The ``(x, y)`` centroid coordinate, calculated by fitting a 2D
quadratic polynomial to the unmasked pixels in the source
segment.
Notes
-----
`~photutils.centroids.centroid_quadratic` is used to calculate
the centroid with ``fit_boxsize=3``.
Because this centroid is based on fitting data, it can fail for
many reasons, returning (np.nan, np.nan):
* quadratic fit failed
* quadratic fit does not have a maximum
* quadratic fit maximum falls outside image
* not enough unmasked data points (6 are required)
Also note that a fit is not performed if the maximum data value
is at the edge of the source segment. In this case, the position
of the maximum pixel will be returned.
"""
origin = np.transpose((self.bbox_xmin, self.bbox_ymin))
return self.cutout_centroid_quad + origin
@lazyproperty
@use_detcat
@as_scalar
def x_centroid_quad(self):
"""
The ``x`` coordinate of the centroid (`centroid_quad`),
calculated by fitting a 2D quadratic polynomial to the unmasked
pixels in the source segment.
"""
if self.isscalar:
x_centroid = self.centroid_quad[0] # scalar array
else:
x_centroid = self.centroid_quad[:, 0]
return x_centroid
@lazyproperty
@use_detcat
@as_scalar
def y_centroid_quad(self):
"""
The ``y`` coordinate of the centroid (`centroid_quad`),
calculated by fitting a 2D quadratic polynomial to the unmasked
pixels in the source segment.
"""
if self.isscalar:
y_centroid = self.centroid_quad[1] # scalar array
else:
y_centroid = self.centroid_quad[:, 1]
return y_centroid
@lazyproperty
@use_detcat
@as_scalar
def sky_centroid(self):
"""
The sky coordinate of the `centroid` within the source segment,
returned as a `~astropy.coordinates.SkyCoord` object.
The output coordinate frame is the same as the input ``wcs``.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.wcs.pixel_to_world(self.x_centroid, self.y_centroid)
@lazyproperty
@use_detcat
@as_scalar
def sky_centroid_icrs(self):
"""
The sky coordinate in the International Celestial Reference
System (ICRS) frame of the `centroid` within the source segment,
returned as a `~astropy.coordinates.SkyCoord` object.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.sky_centroid.icrs
@lazyproperty
@use_detcat
@as_scalar
def sky_centroid_win(self):
"""
The sky coordinate of the "windowed" centroid (`centroid_win`)
within the source segment, returned as a
`~astropy.coordinates.SkyCoord` object.
The output coordinate frame is the same as the input ``wcs``.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.wcs.pixel_to_world(self.x_centroid_win,
self.y_centroid_win)
@lazyproperty
@use_detcat
@as_scalar
def sky_centroid_quad(self):
"""
The sky coordinate of the centroid (`centroid_quad`), calculated
by fitting a 2D quadratic polynomial to the unmasked pixels in
the source segment.
The output coordinate frame is the same as the input ``wcs``.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.wcs.pixel_to_world(self.x_centroid_quad,
self.y_centroid_quad)
@lazyproperty
@use_detcat
def _bbox(self):
"""
The `~photutils.aperture.BoundingBox` of the minimal rectangular
region containing the source segment, always as an iterable.
"""
return [BoundingBox(ixmin=slc[1].start, ixmax=slc[1].stop,
iymin=slc[0].start, iymax=slc[0].stop)
for slc in self._slices_iter]
@lazyproperty
@use_detcat
@as_scalar
def bbox(self):
"""
The `~photutils.aperture.BoundingBox` of the minimal rectangular
region containing the source segment.
Returns a list for multi-source catalogs, or a single
`~photutils.aperture.BoundingBox` for a single-source catalog.
"""
return self._bbox
@lazyproperty
@use_detcat
@as_scalar
def bbox_xmin(self):
"""
The minimum ``x`` pixel index within the minimal bounding box
containing the source segment.
"""
return np.array([slc[1].start for slc in self._slices_iter])
@lazyproperty
@use_detcat
@as_scalar
def bbox_xmax(self):
"""
The maximum ``x`` pixel index within the minimal bounding box
containing the source segment.
Note that this value is inclusive, unlike numpy slice indices.
"""
return np.array([slc[1].stop - 1 for slc in self._slices_iter])
@lazyproperty
@use_detcat
@as_scalar
def bbox_ymin(self):
"""
The minimum ``y`` pixel index within the minimal bounding box
containing the source segment.
"""
return np.array([slc[0].start for slc in self._slices_iter])
@lazyproperty
@use_detcat
@as_scalar
def bbox_ymax(self):
"""
The maximum ``y`` pixel index within the minimal bounding box
containing the source segment.
Note that this value is inclusive, unlike numpy slice indices.
"""
return np.array([slc[0].stop - 1 for slc in self._slices_iter])
@lazyproperty
@use_detcat
def _bbox_corner_ll(self):
"""
Lower-left *outside* pixel corner location (not index).
"""
return np.array([(bbox_.ixmin - 0.5, bbox_.iymin - 0.5)
for bbox_ in self._bbox])
@lazyproperty
@use_detcat
def _bbox_corner_ul(self):
"""
Upper-left *outside* pixel corner location (not index).
"""
return np.array([(bbox_.ixmin - 0.5, bbox_.iymax + 0.5)
for bbox_ in self._bbox])
@lazyproperty
@use_detcat
def _bbox_corner_lr(self):
"""
Lower-right *outside* pixel corner location (not index).
"""
return np.array([(bbox_.ixmax + 0.5, bbox_.iymin - 0.5)
for bbox_ in self._bbox])
@lazyproperty
@use_detcat
def _bbox_corner_ur(self):
"""
Upper-right *outside* pixel corner location (not index).
"""
return np.array([(bbox_.ixmax + 0.5, bbox_.iymax + 0.5)
for bbox_ in self._bbox])
@lazyproperty
@use_detcat
@as_scalar
def sky_bbox_ll(self):
"""
The sky coordinates of the lower-left corner vertex of the
minimal bounding box of the source segment, returned as a
`~astropy.coordinates.SkyCoord` object.
The bounding box encloses all the source segment pixels in their
entirety, thus the vertices are at the pixel *corners*, not
their centers.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.wcs.pixel_to_world(*np.transpose(self._bbox_corner_ll))
@lazyproperty
@use_detcat
@as_scalar
def sky_bbox_ul(self):
"""
The sky coordinates of the upper-left corner vertex of the
minimal bounding box of the source segment, returned as a
`~astropy.coordinates.SkyCoord` object.
The bounding box encloses all the source segment pixels in their
entirety, thus the vertices are at the pixel *corners*, not
their centers.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.wcs.pixel_to_world(*np.transpose(self._bbox_corner_ul))
@lazyproperty
@use_detcat
@as_scalar
def sky_bbox_lr(self):
"""
The sky coordinates of the lower-right corner vertex of the
minimal bounding box of the source segment, returned as a
`~astropy.coordinates.SkyCoord` object.
The bounding box encloses all the source segment pixels in their
entirety, thus the vertices are at the pixel *corners*, not
their centers.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.wcs.pixel_to_world(*np.transpose(self._bbox_corner_lr))
@lazyproperty
@use_detcat
@as_scalar
def sky_bbox_ur(self):
"""
The sky coordinates of the upper-right corner vertex of the
minimal bounding box of the source segment, returned as a
`~astropy.coordinates.SkyCoord` object.
The bounding box encloses all the source segment pixels in their
entirety, thus the vertices are at the pixel *corners*, not
their centers.
`None` if ``wcs`` is not input.
"""
if self.wcs is None:
return self._null_objects
return self.wcs.pixel_to_world(*np.transpose(self._bbox_corner_ur))
@lazyproperty
@as_scalar
def min_value(self):
"""
The minimum pixel value of the ``data`` within the source
segment.
"""
values, _ = self._reduceat(self._data_values, np.minimum)
values -= self._local_background
if self._data_unit is not None:
values <<= self._data_unit
return values
@lazyproperty
@as_scalar
def max_value(self):
"""
The maximum pixel value of the ``data`` within the source
segment.
"""
values, _ = self._reduceat(self._data_values, np.maximum)
values -= self._local_background
if self._data_unit is not None:
values <<= self._data_unit
return values
@lazyproperty
@as_scalar
def cutout_min_value_index(self):
"""
The ``(y, x)`` coordinate, relative to the cutout data, of the
minimum pixel value of the ``data`` within the source segment.
If there are multiple occurrences of the minimum value, only the
first occurrence is returned.
"""
data = self.data_cutout_masked
if self.isscalar:
data = (data,)
idx = []
for arr in data:
if np.all(arr.mask):
idx.append((np.nan, np.nan))
else:
idx.append(np.unravel_index(np.argmin(arr), arr.shape))
return np.array(idx)
@lazyproperty
@as_scalar
def cutout_max_value_index(self):
"""
The ``(y, x)`` coordinate, relative to the cutout data, of the
maximum pixel value of the ``data`` within the source segment.
If there are multiple occurrences of the maximum value, only the
first occurrence is returned.
"""
data = self.data_cutout_masked
if self.isscalar:
data = (data,)
idx = []
for arr in data:
if np.all(arr.mask):
idx.append((np.nan, np.nan))
else:
idx.append(np.unravel_index(np.argmax(arr), arr.shape))
return np.array(idx)
@lazyproperty
@as_scalar
def min_value_index(self):
"""
The ``(y, x)`` coordinate of the minimum pixel value of the
``data`` within the source segment.
If there are multiple occurrences of the minimum value, only the
first occurrence is returned.
"""
index = self.cutout_min_value_index
if self.isscalar:
index = (index,)
out = []
for idx, slc in zip(index, self._slices_iter, strict=True):
out.append((idx[0] + slc[0].start, idx[1] + slc[1].start))
return np.array(out)
@lazyproperty
@as_scalar
def max_value_index(self):
"""
The ``(y, x)`` coordinate of the maximum pixel value of the
``data`` within the source segment.
If there are multiple occurrences of the maximum value, only the
first occurrence is returned.
"""
index = self.cutout_max_value_index
if self.isscalar:
index = (index,)
out = []
for idx, slc in zip(index, self._slices_iter, strict=True):
out.append((idx[0] + slc[0].start, idx[1] + slc[1].start))
return np.array(out)
@lazyproperty
@as_scalar
def min_value_xindex(self):
"""
The ``x`` coordinate of the minimum pixel value of the ``data``
within the source segment.
If there are multiple occurrences of the minimum value, only the
first occurrence is returned.
"""
if self.isscalar:
xidx = self.min_value_index[1]
else:
xidx = self.min_value_index[:, 1]
return xidx
@lazyproperty
@as_scalar
def min_value_yindex(self):
"""
The ``y`` coordinate of the minimum pixel value of the ``data``
within the source segment.
If there are multiple occurrences of the minimum value, only the
first occurrence is returned.
"""
if self.isscalar:
yidx = self.min_value_index[0]
else:
yidx = self.min_value_index[:, 0]
return yidx
@lazyproperty
@as_scalar
def max_value_xindex(self):
"""
The ``x`` coordinate of the maximum pixel value of the ``data``
within the source segment.
If there are multiple occurrences of the maximum value, only the
first occurrence is returned.
"""
if self.isscalar:
xidx = self.max_value_index[1]
else:
xidx = self.max_value_index[:, 1]
return xidx
@lazyproperty
@as_scalar
def max_value_yindex(self):
"""
The ``y`` coordinate of the maximum pixel value of the ``data``
within the source segment.
If there are multiple occurrences of the maximum value, only the
first occurrence is returned.
"""
if self.isscalar:
yidx = self.max_value_index[0]
else:
yidx = self.max_value_index[:, 0]
return yidx
@lazyproperty
@as_scalar
def segment_flux(self):
r"""
The sum of the unmasked ``data`` values within the source
segment.
.. math::
F = \sum_{i \in S} I_i
where :math:`F` is ``segment_flux``, :math:`I_i` is the
background-subtracted ``data``, and :math:`S` are the unmasked
pixels in the source segment.
Non-finite pixel values (NaN and inf) are excluded
(automatically masked).
"""
localbkg = self._local_background
if self.isscalar:
localbkg = localbkg[0]
source_sum, _ = self._reduceat(self._data_values, np.add)
source_sum -= self.area.value * localbkg
if self._data_unit is not None:
source_sum <<= self._data_unit
return source_sum
@lazyproperty
@as_scalar
def segment_flux_err(self):
r"""
The uncertainty of `segment_flux`, propagated from the input
``error`` array.
``segment_flux_err`` is the quadrature sum of the total errors
over the unmasked pixels within the source segment:
.. math::
\Delta F = \sqrt{\sum_{i \in S} \sigma_{\mathrm{tot}, i}^2}
where :math:`\Delta F` is the `segment_flux_err`,
:math:`\sigma_{\mathrm{tot, i}}` are the pixel-wise total errors
(``error``), and :math:`S` are the unmasked pixels in the source
segment.
Pixel values that are masked in the input ``data``, including
any non-finite pixel values (NaN and inf) that are automatically
masked, are also masked in the error array.
"""
if self._error is None:
err = self._null_values
else:
err_sq, _ = self._reduceat(self._error_values, np.add,
transform=np.square)
err = np.sqrt(err_sq)
if self._data_unit is not None:
err <<= self._data_unit
return err
@lazyproperty
@as_scalar
def background_sum(self):
"""
The sum of ``background`` values within the source segment.
Pixel values that are masked in the input ``data``, including
any non-finite pixel values (NaN and inf) that are automatically
masked, are also masked in the background array.
"""
if self._background is None:
bkg_sum = self._null_values
else:
bkg_sum, _ = self._reduceat(
self._background_values, np.add)
if self._data_unit is not None:
bkg_sum <<= self._data_unit
return bkg_sum
@lazyproperty
@as_scalar
def background_mean(self):
"""
The mean of ``background`` values within the source segment.
Pixel values that are masked in the input ``data``, including
any non-finite pixel values (NaN and inf) that are automatically
masked, are also masked in the background array.
"""
if self._background is None:
bkg_mean = self._null_values
else:
bkg_sum, sizes = self._reduceat(
self._background_values, np.add)
bkg_mean = bkg_sum / sizes
if self._data_unit is not None:
bkg_mean <<= self._data_unit
return bkg_mean
@lazyproperty
@as_scalar
def background_centroid(self):
"""
The value of the per-pixel ``background`` at the position of the
isophotal (center-of-mass) `centroid`.
If ``detection_catalog`` is input, then its `centroid` will be
used.
The background values at fractional position values are
determined using bilinear interpolation.
"""
if self._background is None:
bkg = self._null_values
else:
xcen = self._x_centroid
ycen = self._y_centroid
bkg = map_coordinates(self._background, (ycen, xcen), order=1,
mode='nearest')
mask = np.isfinite(xcen) & np.isfinite(ycen)
bkg[~mask] = np.nan
if self._data_unit is not None:
bkg <<= self._data_unit
return bkg
@lazyproperty
@use_detcat
@as_scalar
def segment_area(self):
"""
The total area of the source segment in units of pixels**2.
This area is simply the area of the source segment from the
input ``segmentation_image``. It does not take into account
any data masking (i.e., a ``mask`` input to `SourceCatalog` or
invalid ``data`` values).
"""
areas = []
for label, slices in zip(self.labels, self._slices_iter,
strict=True):
areas.append(np.count_nonzero(
self._segmentation_image[slices] == label))
return np.array(areas) << (u.pix**2)
@lazyproperty
@use_detcat
@as_scalar
def area(self):
"""
The total unmasked area of the source in units of pixels**2.
Note that the source area may be smaller than its `segment_area`
if a mask is input to `SourceCatalog` or if the ``data`` within
the segment contains invalid values (NaN and inf).
"""
areas = np.array([arr.size for arr in self._data_values]).astype(float)
areas[self._all_masked] = np.nan
return areas << (u.pix**2)
@lazyproperty
@use_detcat
@as_scalar
def equivalent_radius(self):
"""
The radius of a circle with the same `area` as the source
segment.
"""
return np.sqrt(self.area / np.pi)
@lazyproperty
@use_detcat
@as_scalar
def perimeter(self):
"""
The perimeter of the source segment, approximated as the total
length of lines connecting the centers of the border pixels
defined by a 4-pixel connectivity.
If any masked pixels make holes within the source segment, then
the perimeter around the inner hole (e.g., an annulus) will also
contribute to the total perimeter.
References
----------
.. [1] K. Benkrid, D. Crookes, and A. Benkrid. "Design and FPGA
Implementation of a Perimeter Estimator". Proceedings of
the Irish Machine Vision and Image Processing Conference,
pp. 51-57 (2000).
"""
size = 34
weights = np.zeros(size, dtype=float)
weights[[5, 7, 15, 17, 25, 27]] = 1.0
weights[[21, 33]] = np.sqrt(2.0)
weights[[13, 23]] = (1 + np.sqrt(2.0)) / 2.0
perimeter = []
for mask in self._cutout_total_masks:
if np.all(mask):
perimeter.append(np.nan)
continue
ny, nx = mask.shape
# Pad source array with zeros (border_value=0)
padded = np.zeros((ny + 2, nx + 2), dtype=np.int8)
padded[1:-1, 1:-1] = ~mask
# Binary erosion with cross footprint (4-connectivity):
# a pixel is eroded if any 4-connected neighbor is 0
p = padded
eroded = (p[1:-1, 1:-1] & p[:-2, 1:-1] & p[2:, 1:-1]
& p[1:-1, :-2] & p[1:-1, 2:])
# Border pixels are source pixels that were eroded away
border = np.zeros((ny + 2, nx + 2), dtype=np.int8)
border[1:-1, 1:-1] = padded[1:-1, 1:-1] & ~eroded
# Convolution with kernel [[10,2,10], [2,1,2], [10,2,10]]
b = border
conv = (10 * b[:-2, :-2] + 2 * b[:-2, 1:-1]
+ 10 * b[:-2, 2:] + 2 * b[1:-1, :-2]
+ b[1:-1, 1:-1] + 2 * b[1:-1, 2:]
+ 10 * b[2:, :-2] + 2 * b[2:, 1:-1]
+ 10 * b[2:, 2:])
hist = np.bincount(conv.ravel(), minlength=size)
perimeter.append(hist[:size] @ weights)
return np.array(perimeter) * u.pix
@lazyproperty
@use_detcat
@as_scalar
def inertia_tensor(self):
"""
The inertia tensor of the source for the rotation around its
center of mass.
"""
moments = self.moments_central
if self.isscalar:
moments = moments[np.newaxis, :]
mu_02 = moments[:, 0, 2]
mu_11 = -moments[:, 1, 1]
mu_20 = moments[:, 2, 0]
tensor = np.array([mu_02, mu_11, mu_11, mu_20]).swapaxes(0, 1)
return tensor.reshape((tensor.shape[0], 2, 2)) * u.pix**2
@lazyproperty
@use_detcat
def _covariance(self):
"""
The covariance matrix of the 2D Gaussian function that has the
same second-order moments as the source, always as an iterable.
"""
moments = self.moments_central
if self.isscalar:
moments = moments[np.newaxis, :]
# Ignore divide-by-zero RuntimeWarning
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
mu_norm = moments / moments[:, 0, 0][:, np.newaxis, np.newaxis]
covar = np.array([mu_norm[:, 0, 2], mu_norm[:, 1, 1],
mu_norm[:, 1, 1], mu_norm[:, 2, 0]]).swapaxes(0, 1)
covar = covar.reshape((covar.shape[0], 2, 2))
# Modify the covariance matrix in the case of "infinitely" thin
# detections. This follows SourceExtractor's prescription of
# incrementally increasing the diagonal elements by 1/12.
delta = 1.0 / 12
delta2 = delta**2
# Ignore RuntimeWarning from NaN values in covar
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
covar_det = np.linalg.det(covar)
# Covariance should be positive semidefinite
idx = np.where(covar_det < 0)[0]
covar[idx] = np.array([[np.nan, np.nan], [np.nan, np.nan]])
idx = np.where(covar_det < delta2)[0]
while idx.size > 0:
covar[idx, 0, 0] += delta
covar[idx, 1, 1] += delta
covar_det = np.linalg.det(covar)
idx = np.where(covar_det < delta2)[0]
return covar
@lazyproperty
@use_detcat
@as_scalar
def covariance(self):
"""
The covariance matrix of the 2D Gaussian function that has the
same second-order moments as the source.
"""
return self._covariance * (u.pix**2)
@lazyproperty
@use_detcat
@as_scalar
def covariance_eigvals(self):
"""
The two eigenvalues of the `covariance` matrix in decreasing
order.
"""
eigvals = np.empty((self.n_labels, 2))
eigvals.fill(np.nan)
# np.linalg.eigvalsh requires finite input values
idx = np.unique(np.where(np.isfinite(self._covariance))[0])
eigvals[idx] = np.linalg.eigvalsh(self._covariance[idx])
# Check for negative variance
# (just in case covariance matrix is not positive semidefinite)
idx2 = np.unique(np.where(eigvals < 0)[0])
eigvals[idx2] = (np.nan, np.nan)
# Sort each eigenvalue pair in descending order
# (eigvalsh returns values in ascending order)
eigvals = np.fliplr(eigvals)
return eigvals * u.pix**2
@lazyproperty
@use_detcat
@as_scalar
def semimajor_axis(self):
"""
The 1-sigma standard deviation along the semimajor axis of the
2D Gaussian function that has the same second-order central
moments as the source.
"""
eigvals = self.covariance_eigvals
if self.isscalar:
eigvals = eigvals[np.newaxis, :]
# This matches SourceExtractor's A parameter
return np.sqrt(eigvals[:, 0])
@lazyproperty
@use_detcat
@as_scalar
def semiminor_axis(self):
"""
The 1-sigma standard deviation along the semiminor axis of the
2D Gaussian function that has the same second-order central
moments as the source.
"""
eigvals = self.covariance_eigvals
if self.isscalar:
eigvals = eigvals[np.newaxis, :]
# This matches SourceExtractor's B parameter
return np.sqrt(eigvals[:, 1])
@lazyproperty
@use_detcat
@as_scalar
def fwhm(self):
r"""
The circularized full width at half maximum (FWHM) of the 2D
Gaussian function that has the same second-order central moments
as the source.
.. math::
\mathrm{FWHM} & = 2 \sqrt{2 \ln(2)} \sqrt{0.5 (a^2 + b^2)}
\\
& = 2 \sqrt{\ln(2) \ (a^2 + b^2)}
where :math:`a` and :math:`b` are the 1-sigma lengths of the
semimajor (`semimajor_axis`) and semiminor (`semiminor_axis`)
axes, respectively.
"""
return 2.0 * np.sqrt(np.log(2.0) * (self.semimajor_axis**2
+ self.semiminor_axis**2))
@lazyproperty
@use_detcat
@as_scalar
def orientation(self):
"""
The angle between the ``x`` axis and the major axis of the 2D
Gaussian function that has the same second-order moments as the
source.
The angle increases in the counter-clockwise direction and
will be in the range [0, 360) degrees.
"""
covar = self._covariance
orient_radians = 0.5 * np.arctan2(2.0 * covar[:, 0, 1],
(covar[:, 0, 0] - covar[:, 1, 1]))
return (np.rad2deg(orient_radians) % 360) << u.deg
@lazyproperty
@use_detcat
@as_scalar
def eccentricity(self):
r"""
The eccentricity of the 2D Gaussian function that has the same
second-order moments as the source.
The eccentricity is the fraction of the distance along the
semimajor axis at which the focus lies.
.. math::
e = \sqrt{1 - \frac{b^2}{a^2}}
where :math:`a` and :math:`b` are the lengths of the semimajor
and semiminor axes, respectively.
"""
semimajor_var, semiminor_var = np.transpose(self.covariance_eigvals)
return np.sqrt(1.0 - (semiminor_var / semimajor_var))
@lazyproperty
@use_detcat
@as_scalar
def elongation(self):
r"""
The ratio of the lengths of the semimajor and semiminor axes.
.. math::
\mathrm{elongation} = \frac{a}{b}
where :math:`a` and :math:`b` are the lengths of the semimajor
and semiminor axes, respectively.
"""
return self.semimajor_axis / self.semiminor_axis
@lazyproperty
@use_detcat
@as_scalar
def ellipticity(self):
r"""
1.0 minus the ratio of the lengths of the semimajor and
semiminor axes.
.. math::
\mathrm{ellipticity} = \frac{a - b}{a} = 1 - \frac{b}{a}
where :math:`a` and :math:`b` are the lengths of the semimajor
and semiminor axes, respectively.
"""
return 1.0 - (self.semiminor_axis / self.semimajor_axis)
@lazyproperty
@use_detcat
@as_scalar
def covariance_xx(self):
r"""
The ``(0, 0)`` element of the `covariance` matrix, representing
:math:`\sigma_x^2`, in units of pixel**2.
"""
return self._covariance[:, 0, 0] * u.pix**2
@lazyproperty
@use_detcat
@as_scalar
def covariance_yy(self):
r"""
The ``(1, 1)`` element of the `covariance` matrix, representing
:math:`\sigma_y^2`, in units of pixel**2.
"""
return self._covariance[:, 1, 1] * u.pix**2
@lazyproperty
@use_detcat
@as_scalar
def covariance_xy(self):
r"""
The ``(0, 1)`` and ``(1, 0)`` elements of the `covariance`
matrix, representing :math:`\sigma_x \sigma_y`, in units of
pixel**2.
"""
return self._covariance[:, 0, 1] * u.pix**2
@lazyproperty
@use_detcat
@as_scalar
def ellipse_cxx(self):
r"""
Coefficient for ``x**2`` in the generalized ellipse equation in
units of pixel**(-2).
The ellipse is defined as
.. math::
cxx (x - \bar{x})^2 + cxy (x - \bar{x}) (y - \bar{y}) +
cyy (y - \bar{y})^2 = R^2
where :math:`R` is a parameter which scales the ellipse (in
units of the axes lengths).
`SourceExtractor`_ reports that the isophotal limit of a source
is well represented by :math:`R \approx 3`.
"""
return ((np.cos(self.orientation) / self.semimajor_axis)**2
+ (np.sin(self.orientation) / self.semiminor_axis)**2)
@lazyproperty
@use_detcat
@as_scalar
def ellipse_cyy(self):
r"""
Coefficient for ``y**2`` in the generalized ellipse equation in
units of pixel**(-2).
The ellipse is defined as
.. math::
cxx (x - \bar{x})^2 + cxy (x - \bar{x}) (y - \bar{y}) +
cyy (y - \bar{y})^2 = R^2
where :math:`R` is a parameter which scales the ellipse (in
units of the axes lengths).
`SourceExtractor`_ reports that the isophotal limit of a source
is well represented by :math:`R \approx 3`.
"""
return ((np.sin(self.orientation) / self.semimajor_axis)**2
+ (np.cos(self.orientation) / self.semiminor_axis)**2)
@lazyproperty
@use_detcat
@as_scalar
def ellipse_cxy(self):
r"""
Coefficient for ``x * y`` in the generalized ellipse equation in
units of pixel**(-2).
The ellipse is defined as
.. math::
cxx (x - \bar{x})^2 + cxy (x - \bar{x}) (y - \bar{y}) +
cyy (y - \bar{y})^2 = R^2
where :math:`R` is a parameter which scales the ellipse (in
units of the axes lengths).
`SourceExtractor`_ reports that the isophotal limit of a source
is well represented by :math:`R \approx 3`.
"""
return (2.0 * np.cos(self.orientation) * np.sin(self.orientation)
* ((1.0 / self.semimajor_axis**2)
- (1.0 / self.semiminor_axis**2)))
@lazyproperty
@use_detcat
@as_scalar
def gini(self):
r"""
The `Gini coefficient
<https://en.wikipedia.org/wiki/Gini_coefficient>`_ of the
source.
The Gini coefficient of the distribution of absolute flux values
is calculated using the prescription from `Lotz et al. 2004
<https://ui.adsabs.harvard.edu/abs/2004AJ....128..163L/abstract>`_
(Eq. 6) as:
.. math::
G = \frac{1}{\overline{|x|} \, n \, (n - 1)}
\sum^{n}_{i} (2i - n - 1) \left | x_i \right |
where :math:`\overline{|x|}` is the mean of the absolute value
of all pixel values :math:`x_i`. If the sum of all pixel values
is zero, the Gini coefficient is zero.
Negative pixel values are used via their absolute value. Invalid
values (NaN and inf) in the input are automatically excluded
from the calculation. If only a single finite pixel remains
after filtering, the Gini coefficient is 0.0.
"""
return np.array([gini_func(arr) for arr in self._data_values])
@lazyproperty
def _local_background_apertures(self):
"""
The `~photutils.aperture.RectangularAnnulus` aperture used to
estimate the local background.
"""
if self.local_bkg_width == 0:
return self._null_objects
apertures = []
for bbox_ in self._bbox:
xpos = 0.5 * (bbox_.ixmin + bbox_.ixmax - 1)
ypos = 0.5 * (bbox_.iymin + bbox_.iymax - 1)
scale = 1.5
width_in = (bbox_.ixmax - bbox_.ixmin) * scale
width_out = width_in + 2 * self.local_bkg_width
height_in = (bbox_.iymax - bbox_.iymin) * scale
height_out = height_in + 2 * self.local_bkg_width
apertures.append(RectangularAnnulus((xpos, ypos), width_in,
width_out, height_out,
h_in=height_in, theta=0.0))
return apertures
@lazyproperty
@use_detcat
@as_scalar
def local_background_aperture(self):
"""
The `~photutils.aperture.RectangularAnnulus` aperture used to
estimate the local background.
Returns a list of apertures for multi-source catalogs, or a
single aperture for a single-source catalog.
"""
return self._local_background_apertures
@lazyproperty
def _local_background(self):
"""
The local background value (per pixel) estimated using a
rectangular annulus aperture around the source.
Pixels are masked where the input ``mask`` is `True`, where the
input ``data`` is non-finite, and within any non-zero pixel
label in the segmentation image.
This property is always an `~numpy.ndarray` without units.
"""
if self.local_bkg_width == 0:
local_bkgs = np.zeros(self.n_labels)
else:
sigma_clip = SigmaClip(sigma=3.0, cenfunc='median', maxiters=20)
bkg_func = SExtractorBackground(sigma_clip=sigma_clip)
bkg_apers = self._local_background_apertures
local_bkgs = []
for aperture in bkg_apers:
aperture_mask = aperture.to_mask(method='center')
slc_lg, slc_sm = aperture_mask.get_overlap_slices(
self._data.shape)
data_cutout = self._data[slc_lg].astype(float, copy=True)
# All non-zero segment labels are masked
segm_mask_cutout = (
self._segmentation_image.data[slc_lg].astype(bool))
if self._mask is None:
mask_cutout = None
else:
mask_cutout = self._mask[slc_lg]
data_mask_cutout = self._make_cutout_data_mask(data_cutout,
mask_cutout)
data_mask_cutout |= segm_mask_cutout
aperweight_cutout = aperture_mask.data[slc_sm]
good_mask = (aperweight_cutout > 0) & ~data_mask_cutout
data_cutout *= aperweight_cutout
data_values = data_cutout[good_mask] # 1D array
# Check not enough unmasked pixels
if len(data_values) < 10:
local_bkgs.append(0.0)
continue
local_bkgs.append(bkg_func(data_values))
local_bkgs = np.array(local_bkgs)
local_bkgs[self._all_masked] = np.nan
return local_bkgs
@lazyproperty
@as_scalar
def local_background(self):
"""
The local background value (per pixel) estimated using a
rectangular annulus aperture around the source.
"""
bkg = self._local_background
if self._data_unit is not None:
bkg <<= self._data_unit
return bkg
def _aperture_to_mask(self, aperture, **kwargs):
"""
Call ``aperture.to_mask()``, but first check that the aperture
bounding box is not larger than the input data to prevent
out-of-memory errors from pathologically large apertures.
The aperture mask is allocated at the full (unclipped) bounding
box size by ``to_mask()``, before ``get_overlap_slices`` clips
it to the data shape. For pathological apertures (e.g., from
huge Kron radii), this allocation can cause out-of-memory
issues.
Returns `None` if the aperture mask would be unreasonably large.
"""
bbox = aperture.bbox
# Limit the aperture mask size to prevent OOM errors
max_size = max(self._data.size, 1_000_000)
if bbox.shape[0] * bbox.shape[1] > max_size:
return None
return aperture.to_mask(**kwargs)
def _make_aperture_data(self, label, x_centroid, y_centroid, aperture_bbox,
local_background, *, make_error=True):
"""
Make cutouts of data, error, and mask arrays for aperture
photometry (e.g., circular or Kron).
Neighboring sources can be included, masked, or corrected based
on the ``aperture_mask_method`` keyword.
"""
# Make cutouts of the data based on the aperture bbox
slc_lg, slc_sm = aperture_bbox.get_overlap_slices(self._data.shape)
if slc_lg is None:
return (None,) * 5
data = self._data[slc_lg].astype(float) - local_background
mask_cutout = None if self._mask is None else self._mask[slc_lg]
data_mask = self._make_cutout_data_mask(data, mask_cutout)
if make_error and self._error is not None:
error = self._error[slc_lg]
else:
error = None
# Calculate cutout centroid position
cutout_xycen = (x_centroid - max(0, aperture_bbox.ixmin),
y_centroid - max(0, aperture_bbox.iymin))
# Mask or correct neighboring sources
if self.aperture_mask_method == 'none':
mask = data_mask
else:
segment_img = self._segmentation_image.data[slc_lg]
segm_mask = np.logical_and(segment_img != label,
segment_img != 0)
if self.aperture_mask_method == 'mask':
mask = data_mask | segm_mask
else:
mask = data_mask
if self.aperture_mask_method == 'correct':
data = _mask_to_mirrored_value(data, segm_mask, cutout_xycen,
mask=mask)
if error is not None:
error = _mask_to_mirrored_value(error, segm_mask, cutout_xycen,
mask=mask)
return data, error, mask, cutout_xycen, slc_sm
def _make_circular_apertures(self, radius):
"""
Make circular aperture for each source.
The aperture for each source will be centered at its
`centroid` position. If a ``detection_catalog`` was input to
`SourceCatalog`, it will be used for the source centroids.
Parameters
----------
radius : float, 1D `~numpy.ndarray`
The radius of the circle in pixels.
Returns
-------
result : list of `~photutils.aperture.CircularAperture`
A list of `~photutils.aperture.CircularAperture` instances.
The aperture will be `None` where the source `centroid`
position is not finite or where the source is completely
masked.
"""
radius = np.broadcast_to(radius, len(self._x_centroid))
if np.any(radius <= 0):
msg = 'radius must be > 0'
raise ValueError(msg)
apertures = []
for (xcen, ycen, radius_, all_masked) in zip(self._x_centroid,
self._y_centroid,
radius,
self._all_masked,
strict=True):
if all_masked or np.any(~np.isfinite((xcen, ycen, radius_))):
apertures.append(None)
continue
apertures.append(CircularAperture((xcen, ycen), r=radius_))
return apertures
[docs]
@as_scalar
def make_circular_apertures(self, radius):
"""
Make circular aperture for each source.
The aperture for each source will be centered at its
`centroid` position. If a ``detection_catalog`` was input to
`SourceCatalog`, then its `centroid` values will be used.
Parameters
----------
radius : float
The radius of the circle in pixels.
Returns
-------
result : `~photutils.aperture.CircularAperture` or \
list of `~photutils.aperture.CircularAperture`
The circular aperture for each source. The aperture will be
`None` where the source `centroid` position is not finite or
where the source is completely masked.
"""
return self._make_circular_apertures(radius)
[docs]
@as_scalar
@deprecated_positional_kwargs(since='3.0', until='4.0')
def plot_circular_apertures(self, radius, ax=None, origin=(0, 0),
**kwargs):
"""
Plot circular apertures for each source on a matplotlib
`~matplotlib.axes.Axes` instance.
The aperture for each source will be centered at its
`centroid` position. If a ``detection_catalog`` was input to
`SourceCatalog`, then its `centroid` values will be used.
An aperture will not be plotted for sources where the source
`centroid` position is not finite or where the source is
completely masked.
Parameters
----------
radius : float
The radius of the circle in pixels.
ax : `matplotlib.axes.Axes` or `None`, optional
The matplotlib axes on which to plot. If `None`, then the
current `~matplotlib.axes.Axes` instance is used.
origin : array_like, optional
The ``(x, y)`` position of the origin of the displayed
image.
**kwargs : dict, optional
Any keyword arguments accepted by
`matplotlib.patches.Patch`.
Returns
-------
patch : `~matplotlib.patches.Patch` or \
list of `~matplotlib.patches.Patch`
The matplotlib patch for each plotted aperture. The patches
can be used, for example, when adding a plot legend.
"""
apertures = self._make_circular_apertures(radius)
patches = []
for aperture in apertures:
if aperture is not None:
aperture.plot(ax=ax, origin=origin, **kwargs)
patches.append(aperture._to_patch(origin=origin, **kwargs))
return patches
[docs]
@deprecated_positional_kwargs(since='3.0', until='4.0')
def circular_photometry(self, radius, name=None, overwrite=False):
"""
Perform circular aperture photometry for each source.
The circular aperture for each source will be centered at its
`centroid` position. If a ``detection_catalog`` was input to
`SourceCatalog`, then its `centroid` values will be used.
See the `SourceCatalog` ``aperture_mask_method`` keyword for
options to mask neighboring sources.
Parameters
----------
radius : float
The radius of the circle in pixels.
name : str or `None`, optional
The prefix name which will be used to define attribute
names for the flux and flux error. The attribute names
``[name]_flux`` and ``[name]_flux_err`` will store the
photometry results. For example, these names can then be
included in the `to_table` ``columns`` keyword list to
output the results in the table.
overwrite : bool, optional
If True, overwrite the attribute ``name`` if it exists.
Returns
-------
flux, flux_err : float or `~numpy.ndarray` of floats
The aperture fluxes and flux errors. NaN will be returned
where the aperture is `None` (e.g., where the source
`centroid` position is not finite or the source is
completely masked).
"""
if radius <= 0:
msg = 'radius must be > 0'
raise ValueError(msg)
apertures = self._make_circular_apertures(radius)
kwargs = self._aperture_mask_kwargs['circ']
flux, flux_err = self._aperture_photometry(apertures,
desc='circular_photometry',
**kwargs)
if self._data_unit is not None:
flux <<= self._data_unit
flux_err <<= self._data_unit
if self.isscalar:
flux = flux[0]
flux_err = flux_err[0]
if name is not None:
flux_name = f'{name}_flux'
flux_err_name = f'{name}_flux_err'
self.add_property(flux_name, flux, overwrite=overwrite)
self.add_property(flux_err_name, flux_err,
overwrite=overwrite)
return flux, flux_err
def _make_elliptical_apertures(self, *, scale=6.0):
"""
Return a list of elliptical apertures based on the scaled
isophotal shape of the sources.
If a ``detection_catalog`` was input to `SourceCatalog`, then
its source `centroid` and shape parameters will be used.
If scale is zero (due to a minimum circular radius set in
``kron_params``) then a circular aperture will be returned with
the minimum circular radius.
Parameters
----------
scale : float or `~numpy.ndarray`, optional
The scale factor to apply to the ellipse major and minor
axes. The default value of 6.0 is roughly two times the
isophotal extent of the source. A `~numpy.ndarray` input
must be a 1D array of length ``n_labels``.
Returns
-------
result : list of `~photutils.aperture.EllipticalAperture`
A list of `~photutils.aperture.EllipticalAperture`
instances. The aperture will be `None` where the source
`centroid` position or elliptical shape parameters are not
finite or where the source is completely masked.
"""
xcen = self._x_centroid
ycen = self._y_centroid
major_size = self.semimajor_axis.value * scale
minor_size = self.semiminor_axis.value * scale
theta = self.orientation.to(u.radian).value
if self.isscalar:
major_size = (major_size,)
minor_size = (minor_size,)
theta = (theta,)
aperture = []
for values in zip(xcen, ycen, major_size, minor_size, theta,
self._all_masked, strict=True):
if values[-1] or np.any(~np.isfinite(values[:-1])):
aperture.append(None)
continue
# kron_radius = 0 -> scale = 0 -> major/minor_size = 0
if values[2] == 0 and values[3] == 0:
aperture.append(CircularAperture((values[0], values[1]),
r=self.kron_params[2]))
continue
(xcen_, ycen_, major_, minor_, theta_) = values[:-1]
aperture.append(EllipticalAperture((xcen_, ycen_), major_, minor_,
theta=theta_))
return aperture
@lazyproperty
@use_detcat
def _measured_kron_radius(self):
r"""
The *unscaled* first-moment Kron radius, always as an array
(without units).
The returned value is the measured Kron radius without applying
any minimum Kron or circular radius.
"""
scale = 6.0
xcen_arr = self._x_centroid
ycen_arr = self._y_centroid
a_arr = self.semimajor_axis.value * scale
b_arr = self.semiminor_axis.value * scale
theta_arr = self.orientation.to(u.radian).value
cxx_arr = self.ellipse_cxx.value
cxy_arr = self.ellipse_cxy.value
cyy_arr = self.ellipse_cyy.value
all_masked = self._all_masked
if self.isscalar:
a_arr = (a_arr,)
b_arr = (b_arr,)
theta_arr = (theta_arr,)
cxx_arr = (cxx_arr,)
cxy_arr = (cxy_arr,)
cyy_arr = (cyy_arr,)
data_full = self._data
data_shape = data_full.shape
mask_full = self._mask
segm_data = self._segmentation_image.data
max_size = max(data_full.size, 1_000_000)
kron_min = self.kron_params[1]
min_circ_radius = (self.kron_params[2]
if len(self.kron_params) == 3 else 0.0)
aperture_mask_method = self.aperture_mask_method
labels = self.labels
if self.progress_bar:
desc = 'kron_radius'
labels = add_progress_bar(labels, desc=desc)
kron_radius = []
for (label, xc, yc, a, b, theta, cxx_, cxy_, cyy_,
masked) in zip(labels, xcen_arr, ycen_arr, a_arr, b_arr,
theta_arr, cxx_arr, cxy_arr, cyy_arr,
all_masked, strict=True):
if masked or not (math.isfinite(xc) and math.isfinite(yc)
and math.isfinite(a) and math.isfinite(b)
and math.isfinite(theta)):
kron_radius.append(np.nan)
continue
# Circular aperture fallback when semimajor/semiminor are
# zero (matching _make_elliptical_apertures behavior)
use_circular = (a == 0 and b == 0)
if use_circular:
if min_circ_radius <= 0:
kron_radius.append(np.nan)
continue
half_w = min_circ_radius
half_h = min_circ_radius
else:
cos_theta = math.cos(theta)
sin_theta = math.sin(theta)
half_w = math.sqrt(a * a * cos_theta * cos_theta
+ b * b * sin_theta * sin_theta)
half_h = math.sqrt(a * a * sin_theta * sin_theta
+ b * b * cos_theta * cos_theta)
# Compute bounding box from ellipse/circle parameters
ixmin = math.floor(xc - half_w + 0.5)
ixmax = math.floor(xc + half_w + 0.5) + 1
iymin = math.floor(yc - half_h + 0.5)
iymax = math.floor(yc + half_h + 0.5) + 1
# OOM guard
if (ixmax - ixmin) * (iymax - iymin) > max_size:
kron_radius.append(np.nan)
continue
# Compute overlap slices with data boundaries
dx_min = max(0, -ixmin)
dy_min = max(0, -iymin)
dx_max = max(0, ixmax - data_shape[1])
dy_max = max(0, iymax - data_shape[0])
lg_xmin = ixmin + dx_min
lg_xmax = ixmax - dx_max
lg_ymin = iymin + dy_min
lg_ymax = iymax - dy_max
if lg_xmin >= lg_xmax or lg_ymin >= lg_ymax:
kron_radius.append(np.nan)
continue
slc_lg = (slice(lg_ymin, lg_ymax), slice(lg_xmin, lg_xmax))
# Cutout data (local background explicitly zero for SE
# agreement)
data = data_full[slc_lg].astype(float)
# Build data mask (non-finite + input mask)
data_mask = ~np.isfinite(data)
if mask_full is not None:
data_mask |= mask_full[slc_lg]
# Mask or correct neighboring sources
if aperture_mask_method != 'none':
seg_cut = segm_data[slc_lg]
segm_mask = (seg_cut != label) & (seg_cut != 0)
if aperture_mask_method == 'mask':
mask = data_mask | segm_mask
else:
mask = data_mask
if aperture_mask_method == 'correct':
cutout_xycen = (xc - max(0, ixmin), yc - max(0, iymin))
data = _mask_to_mirrored_value(data, segm_mask,
cutout_xycen,
mask=mask)
else:
mask = data_mask
# Coordinate arrays (ogrid-style broadcasting avoids
# allocating full 2D meshgrid arrays)
ny, nx = data.shape
xval = np.arange(nx) - (xc - lg_xmin)
yval = np.arange(ny) - (yc - lg_ymin)
yy = yval[:, np.newaxis]
xx = xval[np.newaxis, :]
# Elliptical radius
rr_sq = cxx_ * xx * xx + cxy_ * xx * yy + cyy_ * yy * yy
rr = np.sqrt(np.maximum(rr_sq, 0.0))
# Aperture mask: for method='center', pixels whose center
# falls inside the ellipse (rr <= scale) or circle
if use_circular:
dx = xx
dy = yy
pixel_mask = ((dx * dx + dy * dy)
<= min_circ_radius * min_circ_radius) & ~mask
else:
pixel_mask = (rr <= scale) & ~mask
# Ignore RuntimeWarning for invalid data values
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
flux_numer = np.sum(data[pixel_mask] * rr[pixel_mask])
flux_denom = np.sum(data[pixel_mask])
# Set Kron radius to the minimum Kron radius if numerator or
# denominator is negative
if flux_numer <= 0 or flux_denom <= 0:
kron_radius.append(kron_min)
continue
kron_radius.append(flux_numer / flux_denom)
return np.array(kron_radius)
@as_scalar
def _calc_kron_radius(self, kron_params):
"""
Calculate the *unscaled* first-moment Kron radius, applying any
minimum Kron or circular radius to the measured Kron radius.
Returned as a Quantity array or scalar (if self isscalar) with
pixel units.
"""
kron_radius = self._measured_kron_radius.copy()
# Set values exceeding the measurement aperture scale (6.0)
# to NaN. Such values are unphysical (the Kron radius cannot
# meaningfully exceed the aperture used to measure it) and are
# caused by near-cancellation in the denominator of the Kron
# formula due to outlier pixels or noise.
max_kron_radius = 6.0
kron_radius[kron_radius > max_kron_radius] = np.nan
# Set minimum (unscaled) kron radius
kron_radius[kron_radius < kron_params[1]] = kron_params[1]
# Check for minimum circular radius
if len(kron_params) == 3:
semimajor_axis = self.semimajor_axis.value
semiminor_axis = self.semiminor_axis.value
circ_radius = (kron_params[0] * kron_radius
* np.sqrt(semimajor_axis * semiminor_axis))
kron_radius[circ_radius <= kron_params[2]] = 0.0
return kron_radius << u.pix
@lazyproperty
@use_detcat
@as_scalar
def kron_radius(self):
r"""
The *unscaled* first-moment Kron radius.
The *unscaled* first-moment Kron radius is given by:
.. math::
r_k = \frac{\sum_{i \in A} \ r_i I_i}{\sum_{i \in A} I_i}
where :math:`I_i` are the data values and the sum is over
pixels in an elliptical aperture whose axes are defined by
six times the semimajor (`semimajor_axis`) and semiminor
axes (`semiminor_axis`) at the calculated `orientation` (all
properties derived from the central image moments of the
source). :math:`r_i` is the elliptical "radius" to the pixel
given by:
.. math::
r_i^2 = cxx (x_i - \bar{x})^2 +
cxy (x_i - \bar{x})(y_i - \bar{y}) +
cyy (y_i - \bar{y})^2
where :math:`\bar{x}` and :math:`\bar{y}` represent the source
`centroid` and the coefficients are based on image moments
(`ellipse_cxx`, `ellipse_cxy`, and `ellipse_cyy`).
The `kron_radius` value is the unscaled moment value. The
minimum unscaled radius can be set using the second element of
the `SourceCatalog` ``kron_params`` keyword. If the measured
unscaled Kron radius exceeds 6.0 (the measurement aperture
scale factor), ``np.nan`` will be returned. Such values are
unphysical, typically caused by near-cancellation in the
denominator of the Kron formula due to outlier pixels or noise.
If either the numerator or denominator above is less than
or equal to 0, then the minimum unscaled Kron radius
(``kron_params[1]``) will be used.
The Kron aperture is calculated for each source using its shape
parameters, `kron_radius`, and the ``kron_params`` scaling and
minimum values input into `SourceCatalog`. The Kron aperture is
used to compute the Kron photometry.
If ``kron_params[0]`` * `kron_radius` * sqrt(`semimajor_axis` *
`semiminor_axis`) is less than or equal to the minimum circular
radius (``kron_params[2]``), then the Kron radius will be set to
zero and the Kron aperture will be a circle with this minimum
radius.
If the source is completely masked, then ``np.nan`` will be
returned for both the Kron radius and Kron flux (the Kron
aperture will be `None`).
If a ``detection_catalog`` was input to `SourceCatalog`, then
its ``kron_radius`` will be returned.
See the `SourceCatalog` ``aperture_mask_method`` keyword for
options to mask neighboring sources.
"""
return self._calc_kron_radius(self.kron_params)
def _make_kron_apertures(self, kron_params):
"""
Make Kron apertures for each source, always returned as a list.
"""
# NOTE: if kron_radius = NaN, scale = NaN and kron_aperture = None
kron_radius = self._calc_kron_radius(kron_params)
scale = kron_radius.value * kron_params[0]
return self._make_elliptical_apertures(scale=scale)
@lazyproperty
@use_detcat
@as_scalar
def kron_aperture(self):
r"""
The elliptical (or circular) Kron aperture.
The Kron aperture is calculated for each source using its shape
parameters, `kron_radius`, and the ``kron_params`` scaling and
minimum values input into `SourceCatalog`. The Kron aperture is
used to compute the Kron photometry.
If ``kron_params[0]`` * `kron_radius` * sqrt(`semimajor_axis` *
`semiminor_axis`) is less than or equal to the minimum circular
radius (``kron_params[2]``), then the Kron aperture will be a
circle with this minimum radius.
The aperture will be `None` where the source `centroid` position
or elliptical shape parameters are not finite or where the
source is completely masked.
If a ``detection_catalog`` was input to `SourceCatalog`, then
its ``kron_aperture`` will be returned.
Returns a list of apertures for multi-source catalogs, or a
single aperture for a single-source catalog.
"""
return self._make_kron_apertures(self.kron_params)
[docs]
@as_scalar
@deprecated_positional_kwargs(since='3.0', until='4.0')
def make_kron_apertures(self, kron_params=None):
"""
Make Kron apertures for each source.
The aperture for each source will be centered at its
`centroid` position. If a ``detection_catalog`` was input to
`SourceCatalog`, then its `centroid` values will be used.
Note that changing ``kron_params`` from the values
input into `SourceCatalog` does not change the Kron
apertures (`kron_aperture`) and photometry (`kron_flux` and
`kron_flux_err`) in the source catalog. This method should
be used only to explore alternative ``kron_params`` with a
detection image.
Parameters
----------
kron_params : list of 2 or 3 floats or `None`, optional
A list of parameters used to determine the Kron aperture.
The first item is the scaling parameter of the unscaled
Kron radius and the second item represents the minimum
value for the unscaled Kron radius in pixels. The optional
third item is the minimum circular radius in pixels. If
``kron_params[0]`` * `kron_radius` * sqrt(`semimajor_axis`
* `semiminor_axis`) is less than or equal to this radius,
then the Kron aperture will be a circle with this minimum
radius. If `None`, then the ``kron_params`` input into
`SourceCatalog` will be used (the apertures will be the same
as those in `kron_aperture`).
Returns
-------
result : `~photutils.aperture.PixelAperture` \
or list of `~photutils.aperture.PixelAperture`
The Kron apertures for each source. Each aperture will
either be a `~photutils.aperture.EllipticalAperture`,
`~photutils.aperture.CircularAperture`, or `None`. The
aperture will be `None` where the source `centroid` position
or elliptical shape parameters are not finite or where the
source is completely masked.
"""
if kron_params is None:
return self.kron_aperture
return self._make_kron_apertures(kron_params)
[docs]
@as_scalar
@deprecated_positional_kwargs(since='3.0', until='4.0')
def plot_kron_apertures(self, kron_params=None, ax=None, origin=(0, 0),
**kwargs):
"""
Plot Kron apertures for each source on a matplotlib
`~matplotlib.axes.Axes` instance.
The aperture for each source will be centered at its
`centroid` position. If a ``detection_catalog`` was input to
`SourceCatalog`, then its `centroid` values will be used.
An aperture will not be plotted for sources where the source
`centroid` position or elliptical shape parameters are not
finite or where the source is completely masked.
Note that changing ``kron_params`` from the values
input into `SourceCatalog` does not change the Kron
apertures (`kron_aperture`) and photometry (`kron_flux` and
`kron_flux_err`) in the source catalog. This method should be
used only to visualize/explore alternative ``kron_params`` with
a detection image.
Parameters
----------
kron_params : list of 2 or 3 floats or `None`, optional
A list of parameters used to determine the Kron aperture.
The first item is the scaling parameter of the unscaled
Kron radius and the second item represents the minimum
value for the unscaled Kron radius in pixels. The optional
third item is the minimum circular radius in pixels. If
``kron_params[0]`` * `kron_radius` * sqrt(`semimajor_axis`
* `semiminor_axis`) is less than or equal to this radius,
then the Kron aperture will be a circle with this minimum
radius. If `None`, then the ``kron_params`` input into
`SourceCatalog` will be used (the apertures will be the same
as those in `kron_aperture`).
ax : `matplotlib.axes.Axes` or `None`, optional
The matplotlib axes on which to plot. If `None`, then the
current `~matplotlib.axes.Axes` instance is used.
origin : array_like, optional
The ``(x, y)`` position of the origin of the displayed
image.
**kwargs : dict, optional
Any keyword arguments accepted by
`matplotlib.patches.Patch`.
Returns
-------
patch : list of `~matplotlib.patches.Patch`
A list of matplotlib patches for the plotted aperture. The
patches can be used, for example, when adding a plot legend.
"""
if kron_params is None:
apertures = self.kron_aperture
if self.isscalar:
apertures = (apertures,)
else:
apertures = self._make_kron_apertures(kron_params)
patches = []
for aperture in apertures:
if aperture is not None:
aperture.plot(ax=ax, origin=origin, **kwargs)
patches.append(aperture._to_patch(origin=origin, **kwargs))
return patches
def _aperture_photometry(self, apertures, *, desc='', **kwargs):
"""
Perform aperture photometry on cutouts of the data and optional
error arrays.
The appropriate ``aperture_mask_method`` is applied to the
cutouts to handle neighboring sources.
Parameters
----------
apertures : list of `PixelAperture`
A list of the apertures.
desc : str, optional
The description displayed before the progress bar.
**kwargs : dict, optional
Additional keyword arguments passed to the aperture
``to_mask`` method.
Returns
-------
flux, flux_err : 1D `~numpy.ndaray`
The flux and flux error arrays.
"""
labels = self.labels
if self.progress_bar:
labels = add_progress_bar(labels, desc=desc)
flux = []
flux_err = []
for label, aperture, bkg in zip(labels, apertures,
self._local_background, strict=True):
# Return NaN for completely masked sources or sources where
# the centroid is not finite
if aperture is None:
flux.append(np.nan)
flux_err.append(np.nan)
continue
xcen, ycen = aperture.positions
aperture_mask = self._aperture_to_mask(aperture, **kwargs)
if aperture_mask is None:
flux.append(np.nan)
flux_err.append(np.nan)
continue
# Prepare cutouts of the data based on the aperture size
data, error, mask, _, slc_sm = self._make_aperture_data(
label, xcen, ycen, aperture_mask.bbox, bkg)
aperture_weights = aperture_mask.data[slc_sm]
pixel_mask = (aperture_weights > 0) & ~mask # good pixels
# Ignore RuntimeWarning for invalid data or error values
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
values = (aperture_weights * data)[pixel_mask]
flux_ = np.nan if values.shape == (0,) else np.sum(values)
flux.append(flux_)
if error is None:
flux_err_ = np.nan
else:
values = (aperture_weights * error**2)[pixel_mask]
if values.shape == (0,):
flux_err_ = np.nan
else:
flux_err_ = np.sqrt(np.sum(values))
flux_err.append(flux_err_)
flux = np.array(flux)
flux_err = np.array(flux_err)
return flux, flux_err
def _calc_kron_photometry(self, *, kron_params=None):
"""
Calculate the flux and flux error in the Kron aperture (without
units).
See the `SourceCatalog` ``aperture_mask_method`` keyword for
options to mask neighboring sources.
If the Kron aperture is `None`, then ``np.nan`` will be
returned.
If ``detection_catalog`` is input, then its `centroid` values
will be used.
Returns
-------
kron_flux, kron_flux_err : tuple of `~numpy.ndarray`
The Kron flux and flux error.
"""
if kron_params is None:
kron_aperture = self.kron_aperture
if self.isscalar:
kron_aperture = (kron_aperture,)
else:
kron_params = self._validate_kron_params(kron_params)
kron_aperture = self._make_kron_apertures(kron_params)
labels = self.labels
if self.progress_bar:
labels = add_progress_bar(labels, desc='kron_photometry')
_floor = math.floor
max_size = max(self._data.size, 1_000_000)
flux = []
flux_err = []
for label, aperture, bkg in zip(labels, kron_aperture,
self._local_background, strict=True):
if aperture is None:
flux.append(np.nan)
flux_err.append(np.nan)
continue
xcen, ycen = aperture.positions
# Compute the aperture mask directly, bypassing the
# aperture's to_mask() method and ApertureMask/BoundingBox
# property overhead.
if isinstance(aperture, CircularAperture):
r = aperture.r
ixmin = _floor(xcen - r + 0.5)
ixmax = _floor(xcen + r + 1.5)
iymin = _floor(ycen - r + 0.5)
iymax = _floor(ycen + r + 1.5)
nx = ixmax - ixmin
ny = iymax - iymin
if nx * ny > max_size:
flux.append(np.nan)
flux_err.append(np.nan)
continue
edges = (ixmin - 0.5 - xcen, ixmax - 0.5 - xcen,
iymin - 0.5 - ycen, iymax - 0.5 - ycen)
mask_data = circular_overlap_grid(
edges[0], edges[1], edges[2], edges[3],
nx, ny, r, 1, 1)
else:
a = aperture.a
b = aperture.b
theta_val = aperture.theta
theta_rad = (theta_val.to(u.radian).value
if hasattr(theta_val, 'to')
else float(theta_val))
cos_t = math.cos(theta_rad)
sin_t = math.sin(theta_rad)
x_ext = math.sqrt((a * cos_t) ** 2 + (b * sin_t) ** 2)
y_ext = math.sqrt((a * sin_t) ** 2 + (b * cos_t) ** 2)
ixmin = _floor(xcen - x_ext + 0.5)
ixmax = _floor(xcen + x_ext + 1.5)
iymin = _floor(ycen - y_ext + 0.5)
iymax = _floor(ycen + y_ext + 1.5)
nx = ixmax - ixmin
ny = iymax - iymin
if nx * ny > max_size:
flux.append(np.nan)
flux_err.append(np.nan)
continue
edges = (ixmin - 0.5 - xcen, ixmax - 0.5 - xcen,
iymin - 0.5 - ycen, iymax - 0.5 - ycen)
mask_data = elliptical_overlap_grid(
edges[0], edges[1], edges[2], edges[3],
nx, ny, a, b, theta_rad, 1, 1)
bbox = BoundingBox(ixmin, ixmax, iymin, iymax)
data, error, mask, _, slc_sm = self._make_aperture_data(
label, xcen, ycen, bbox, bkg)
if data is None:
flux.append(np.nan)
flux_err.append(np.nan)
continue
aperture_weights = mask_data[slc_sm]
pixel_mask = (aperture_weights > 0) & ~mask
with warnings.catch_warnings():
warnings.simplefilter('ignore', RuntimeWarning)
values = (aperture_weights * data)[pixel_mask]
flux_ = np.nan if values.shape == (0,) else np.sum(values)
flux.append(flux_)
if error is None:
flux_err_ = np.nan
else:
values = (aperture_weights * error ** 2)[pixel_mask]
if values.shape == (0,):
flux_err_ = np.nan
else:
flux_err_ = np.sqrt(np.sum(values))
flux_err.append(flux_err_)
flux = np.array(flux)
flux_err = np.array(flux_err)
return flux, flux_err
[docs]
@deprecated_positional_kwargs(since='3.0', until='4.0')
def kron_photometry(self, kron_params, name=None, overwrite=False):
"""
Perform photometry for each source using an elliptical Kron
aperture.
This method can be used to calculate the Kron photometry using
alternate ``kron_params`` (e.g., different scalings of the Kron
radius).
See the `SourceCatalog` ``aperture_mask_method`` keyword for
options to mask neighboring sources.
Parameters
----------
kron_params : list of 2 or 3 floats, optional
A list of parameters used to determine the Kron aperture.
The first item is the scaling parameter of the unscaled
Kron radius and the second item represents the minimum
value for the unscaled Kron radius in pixels. The optional
third item is the minimum circular radius in pixels. If
``kron_params[0]`` * `kron_radius` * sqrt(`semimajor_axis`
* `semiminor_axis`) is less than or equal to this radius,
then the Kron aperture will be a circle with this minimum
radius.
name : str or `None`, optional
The prefix name which will be used to define attribute
names for the Kron flux and flux error. The attribute
names ``[name]_flux`` and ``[name]_flux_err`` will store
the photometry results. For example, these names can then
be included in the `to_table` ``columns`` keyword list to
output the results in the table.
overwrite : bool, optional
If True, overwrite the attribute ``name`` if it exists.
Returns
-------
flux, flux_err : float or `~numpy.ndarray` of floats
The aperture fluxes and flux errors. NaN will be returned
where the aperture is `None` (e.g., where the source
`centroid` position or elliptical shape parameters are not
finite or where the source is completely masked).
"""
kron_flux, kron_flux_err = self._calc_kron_photometry(
kron_params=kron_params)
if self._data_unit is not None:
kron_flux <<= self._data_unit
kron_flux_err <<= self._data_unit
if self.isscalar:
kron_flux = kron_flux[0]
kron_flux_err = kron_flux_err[0]
if name is not None:
flux_name = f'{name}_flux'
flux_err_name = f'{name}_flux_err'
self.add_property(flux_name, kron_flux, overwrite=overwrite)
self.add_property(flux_err_name, kron_flux_err,
overwrite=overwrite)
return kron_flux, kron_flux_err
@lazyproperty
def _kron_photometry(self):
"""
The flux and flux error in the Kron aperture (without units).
See the `SourceCatalog` ``aperture_mask_method`` keyword for
options to mask neighboring sources.
If the Kron aperture is `None`, then ``np.nan`` will be
returned. This will occur where the source `centroid` position
or elliptical shape parameters are not finite or where the
source is completely masked.
"""
return np.transpose(self._calc_kron_photometry(kron_params=None))
@lazyproperty
@as_scalar
def kron_flux(self):
"""
The flux in the Kron aperture.
See the `SourceCatalog` ``aperture_mask_method`` keyword for
options to mask neighboring sources.
If the Kron aperture is `None`, then ``np.nan`` will be
returned. This will occur where the source `centroid` position
or elliptical shape parameters are not finite or where the
source is completely masked.
"""
kron_flux = self._kron_photometry[:, 0]
if self._data_unit is not None:
kron_flux <<= self._data_unit
return kron_flux
@lazyproperty
@as_scalar
def kron_flux_err(self):
"""
The flux error in the Kron aperture.
See the `SourceCatalog` ``aperture_mask_method`` keyword for
options to mask neighboring sources.
If the Kron aperture is `None`, then ``np.nan`` will be
returned. This will occur where the source `centroid` position
or elliptical shape parameters are not finite or where the
source is completely masked.
"""
kron_flux_err = self._kron_photometry[:, 1]
if self._data_unit is not None:
kron_flux_err <<= self._data_unit
return kron_flux_err
@lazyproperty
@use_detcat
def _max_circular_kron_radius(self):
"""
The maximum circular Kron radius used as the upper limit of
``flux_radius``.
"""
semimajor_sig = self.semimajor_axis.value
kron_radius = self.kron_radius.value
radius = semimajor_sig * kron_radius * self.kron_params[0]
mask = radius == 0
if np.any(mask):
radius[mask] = self.kron_params[2]
if self.isscalar:
radius = np.array([radius])
return radius
@staticmethod
def _flux_radius_fcn(radius, clean_data, grid_params, normflux):
"""
Function whose root is found to compute the flux_radius.
Uses ``circular_overlap_grid`` directly on pre-computed cutout
data (with masked pixels zeroed) to avoid per-call aperture
object overhead.
"""
xmin_e, xmax_e, ymin_e, ymax_e, nx, ny, exact, subpx = grid_params
weights = circular_overlap_grid(xmin_e, xmax_e, ymin_e, ymax_e,
nx, ny, radius, exact, subpx)
flux = np.sum(clean_data * weights)
return 1.0 - (flux / normflux)
@lazyproperty
@use_detcat
def _flux_radius_optimizer_args(self):
kron_flux = self._kron_photometry[:, 0] # unitless
max_radius = self._max_circular_kron_radius
kwargs = self._aperture_mask_kwargs['flux_radius']
# Translate mask method keywords to circular_overlap_grid
# parameters once
method = kwargs.get('method', 'exact')
if method == 'exact':
use_exact = 1
subpixels = 1
elif method == 'center':
use_exact = 0
subpixels = 1
else: # 'subpixel'
use_exact = 0
subpixels = kwargs.get('subpixels', 5)
# Pre-fetch arrays used inside the loop
data_arr = self._data
mask_arr = self._mask
segm_data = self._segmentation_image.data
data_shape = data_arr.shape
aperture_mask_method = self.aperture_mask_method
max_aper_size = max(data_arr.size, 1_000_000)
labels = self.labels
if self.progress_bar:
desc = 'flux_radius prep'
labels = add_progress_bar(labels, desc=desc)
args = []
for label, xcen, ycen, kronflux, bkg, max_radius_ in zip(
labels, self._x_centroid, self._y_centroid,
kron_flux, self._local_background, max_radius, strict=True):
if (np.any(~np.isfinite((xcen, ycen, kronflux, max_radius_)))
or kronflux == 0):
args.append(None)
continue
# Compute the bounding box for the max-radius aperture
# inline, replacing CircularAperture + _aperture_to_mask +
# _make_aperture_data
ixmin = math.floor(xcen - max_radius_ + 0.5)
ixmax = math.ceil(xcen + max_radius_ + 0.5)
iymin = math.floor(ycen - max_radius_ + 0.5)
iymax = math.ceil(ycen + max_radius_ + 0.5)
# OOM guard (same logic as _aperture_to_mask)
bbox_ny = iymax - iymin
bbox_nx = ixmax - ixmin
if bbox_ny * bbox_nx > max_aper_size:
args.append(None)
continue
# Clip to data boundaries
data_ymin = max(0, iymin)
data_ymax = min(data_shape[0], iymax)
data_xmin = max(0, ixmin)
data_xmax = min(data_shape[1], ixmax)
if data_ymin >= data_ymax or data_xmin >= data_xmax:
args.append(None)
continue
slc_lg = (slice(data_ymin, data_ymax), slice(data_xmin, data_xmax))
cutout_data = data_arr[slc_lg].astype(float) - bkg
# Build data mask (non-finite + user mask)
data_mask = ~np.isfinite(cutout_data)
if mask_arr is not None:
data_mask |= mask_arr[slc_lg]
# Cutout centroid position
cutout_xcen = xcen - data_xmin
cutout_ycen = ycen - data_ymin
# Handle neighboring sources
if aperture_mask_method != 'none':
seg_cut = segm_data[slc_lg]
segm_mask = (seg_cut != label) & (seg_cut != 0)
if aperture_mask_method == 'mask':
data_mask = data_mask | segm_mask
elif aperture_mask_method == 'correct':
cutout_data = _mask_to_mirrored_value(
cutout_data, segm_mask,
(cutout_xcen, cutout_ycen), mask=data_mask)
# Pre-zero masked pixels so the root-finding function can
# use a simple sum without masking
clean_data = cutout_data.copy()
clean_data[data_mask] = 0.0
# Pre-compute grid parameters for circular_overlap_grid
ny, nx = clean_data.shape
xmin_edge = -0.5 - cutout_xcen
xmax_edge = nx - 0.5 - cutout_xcen
ymin_edge = -0.5 - cutout_ycen
ymax_edge = ny - 0.5 - cutout_ycen
grid_params = (xmin_edge, xmax_edge, ymin_edge, ymax_edge,
nx, ny, use_exact, subpixels)
args.append([clean_data, grid_params, kronflux, max_radius_])
return args
[docs]
@as_scalar
@deprecated_positional_kwargs(since='3.0', until='4.0')
def flux_radius(self, fraction, name=None, overwrite=False):
"""
Calculate the circular radius that encloses the specified
fraction of the Kron flux.
To estimate the half-light radius, use ``fraction = 0.5``.
Parameters
----------
fraction : float
The fraction of the Kron flux at which to find the circular
radius.
name : str or `None`, optional
The attribute name which will be assigned to the value
of the output array. For example, this name can then be
included in the `to_table` ``columns`` keyword list to
output the results in the table.
overwrite : bool, optional
If True, overwrite the attribute ``name`` if it exists.
Returns
-------
radius : 1D `~numpy.ndarray`
The circular radius that encloses the specified fraction of
the Kron flux. NaN is returned where no solution was found
or where the Kron flux is zero or non-finite.
"""
if fraction <= 0 or fraction > 1:
msg = 'fraction must be > 0 and <= 1'
raise ValueError(msg)
# Return cached result if available
if fraction in self._flux_radius_cache:
result = self._flux_radius_cache[fraction]
if name is not None:
self.add_property(name, result, overwrite=overwrite)
return result
args = self._flux_radius_optimizer_args
if self.progress_bar:
desc = 'flux_radius'
args = add_progress_bar(args, desc=desc)
radius = []
for flux_radius_args in args:
if flux_radius_args is None:
radius.append(np.nan)
continue
clean_data, grid_params, kronflux, max_radius = flux_radius_args
normflux = kronflux * fraction
args = (clean_data, grid_params, normflux)
# Try to find the root of self._flux_radius_func, which
# is bracketed by a min and max radius. A ValueError is
# raised if the bracket points do not have different signs,
# indicating no solution or multiple solutions (e.g., a
# multi-valued function). This can happen when at some
# radius, flux starts decreasing with increasing radius (due
# to negative data values), resulting in multiple possible
# solutions. If no solution is found, we iteratively
# decrease the max radius to narrow the bracket range until
# the root is found. If max radius drops below the min
# radius (0.1), then no solution is possible and NaN will be
# returned as the result.
found = False
min_radius = 0.1
max_radius_delta = 0.1 * max_radius
while max_radius > min_radius and found is False:
try:
bracket = [min_radius, max_radius]
result = root_scalar(self._flux_radius_fcn, args=args,
bracket=bracket, method='brentq')
result = result.root
found = True
except ValueError:
# ValueError is raised if the bracket points do not
# have different signs
max_radius -= max_radius_delta
# No solution found between min_radius and max_radius
if found is False:
result = np.nan
radius.append(result)
result = np.array(radius) << u.pix
self._flux_radius_cache[fraction] = result
if name is not None:
self.add_property(name, result, overwrite=overwrite)
return result
[docs]
@as_scalar
def make_cutouts(self, shape, *, array=None, mode='partial',
fill_value=np.nan):
"""
Make cutout arrays for each source.
The cutout for each source will be centered at its
`centroid` position. If a ``detection_catalog`` was input to
`SourceCatalog`, then its `centroid` values will be used.
Parameters
----------
shape : 2-tuple
The cutout shape along each axis in ``(ny, nx)`` order.
array : `None` or 2D `~numpy.ndarray`
A 2D array with the same shape as the ``data`` array input
to `~photutils.segmentation.SourceCatalog`. If `None` then
the ``data`` array will be used. If any cutout arrays
are not fully contained within the ``array`` array and
``mode='partial'`` with ``fill_value=np.nan``, then the
input ``array`` must have a float data type.
mode : {'partial', 'trim'}, optional
The mode used for extracting the cutout array. In
``'partial'`` mode, positions in the cutout array that
do not overlap with the large array will be filled with
``fill_value``. In ``'trim'`` mode, only the overlapping
elements are returned, thus the resulting small array may be
smaller than the requested ``shape``.
fill_value : number, optional
If ``mode='partial'``, the value to fill pixels in the
extracted cutout array that do not overlap with the input
``array_large``. ``fill_value`` will be changed to have
the same ``dtype`` as the ``array_large`` array, with
one exception. If ``array_large`` has integer type and
``fill_value`` is ``np.nan``, then a `ValueError` will be
raised.
Returns
-------
cutouts : `~photutils.utils.CutoutImage` \
or list of `~photutils.utils.CutoutImage`
The `~photutils.utils.CutoutImage` for each source. The
cutout will be `None` where the source `centroid` position
is not finite or where the source is completely masked.
"""
if array is None:
array = self._data
elif array.shape != self._data.shape:
msg = 'array must have the same shape as data'
raise ValueError(msg)
if mode not in ('partial', 'trim'):
msg = "mode must be 'partial' or 'trim'"
raise ValueError(msg)
cutouts = []
for (xcen, ycen, all_masked) in zip(self._x_centroid,
self._y_centroid,
self._all_masked, strict=True):
if all_masked or np.any(~np.isfinite((xcen, ycen))):
cutouts.append(None)
continue
cutouts.append(CutoutImage(array, (ycen, xcen), shape,
mode=mode, fill_value=fill_value))
return cutouts
