Denoising layer of DC Net

[nducros]

https://spyrit.readthedocs.io/en/master/_autosummary/spyrit.core.recon.Denoise_layer.html#spyrit.core.recon.Denoise_layer

• Equation does not display well
• Notations not consistent (\sigma vs. A=)

[romainphan]

corrected

6628dee

[nducros]

The doc is confusing as the denoise layer only computes the filter (a / a + b), which is designed to be multiplied to some input measurement vector. The usage of the filter is only visible here:

spyrit/spyrit/core/recon.py

Line 191 in aa17397

y1 = torch.mul(self.denoi(var), x)

The design of the class seems awkward, as the forward function should actually compute what the doc says, i.e., (a / a + b) input

[romainphan]

#228