CWT

class gammapy.detect.CWT(kernels, max_iter=10, tol=1e-05, significance_threshold=3.0, significance_island_threshold=None, remove_isolated=True, keep_history=False)[source]

Bases: object

Continuous wavelet transform.

TODO: describe algorithm (modify the words below)

Depending on their spectral index, sources won’t have the same characteristic scale. Therefore to detect sources, we need to compute the wavelet transform at several scales in order to search for various PSF sizes. Then for each scale, the wavelet transform values under the given significance threshold are rejected. This gives us a multiscale support. Then, using the reconstruction by continuous wavelet packets, we obtain a filtered image yielding the detected sources. To compute the threshold image for a given scale a, the standard EGRET diffuse background model to which was added the flux of the extragalactic background, and the exposure for the considered energy range were used.

Parameters:
kernels : CWTKernels

Kernels for the algorithm.

max_iter : int, optional (default 10)

The maximum number of iterations of the CWT algorithm.

tol : float, optional (default 1e-5)

Tolerance for stopping criterion.

significance_threshold : float, optional (default 3.0)

Measure of statistical significance.

significance_island_threshold : float, optional (default None)

Measure is used for cleaning of isolated pixel islands that are slightly above significance_threshold.

remove_isolated : boolean, optional (default True)

If True, isolated pixels will be removed.

keep_history : boolean, optional (default False)

Save cwt data from all the iterations.

References

R. Terrier et al (2001) “Wavelet analysis of EGRET data” See https://ui.adsabs.harvard.edu/abs/2001ICRC….7.2923T

Methods Summary

analyze(self, data) Run iterative filter peak algorithm.

Methods Documentation

analyze(self, data)[source]

Run iterative filter peak algorithm.

The algorithm modifies the original data.

Parameters:
data : CWTData

Input images.