centroid_epsf(data, mask=None, oversampling=4, shift_val=0.5)¶
Deprecated since version 1.2: The centroid_epsf function is deprecated and may be removed in a future version.
Calculate centering shift of data using pixel symmetry, as described by Anderson and King (2000; PASP 112, 1360) in their ePSF-fitting algorithm.
Calculate the shift of a 2-dimensional symmetric image based on the asymmetry between f(x, N) and f(x, -N), along with the differential df/dy(x, shift_val) and df/dy(x, -shift_val). Non-finite values (e.g., NaN or inf) in the
dataarray are automatically masked.
The input n-dimensional array.
- maskarray_like (bool), optional
A boolean mask, with the same shape as
data, where a
Truevalue indicates the corresponding element of
- oversamplingint or tuple of two int, optional
Oversampling factors of pixel indices. If
oversamplingis a scalar this is treated as both x and y directions having the same oversampling factor. Otherwise it is treated as
- shift_valfloat, optional
The undersampled value at which to compute the shifts. Default is half a pixel. It must be a strictly positive number.
- centroidtuple of floats
The (x, y) coordinates of the centroid in pixel order.