In the last paragraph of section 4, you say "Noise is reduced using spatial and temporal
correlations in preprocessing by VBM4D."
So I want to confirm that the preprocessing include Bayer to raw RGB conversion, black level subtraction, binning, global digital gain and VBM4D right?
And if preprocessing including VBM4D denoising, why not just let the network learn denoising instead doing it in the preprocessing step
In the last paragraph of section 4, you say "Noise is reduced using spatial and temporal correlations in preprocessing by VBM4D." So I want to confirm that the preprocessing include Bayer to raw RGB conversion, black level subtraction, binning, global digital gain and VBM4D right? And if preprocessing including VBM4D denoising, why not just let the network learn denoising instead doing it in the preprocessing step