signal analysis question

[toto tata]

hello,

I have a datacube of dimensions 512x512x256 pixels, that I need to analyse. The data represent 256 2D images of the same object in different frequencies.

The aim is to obtain an estimation of the object, deconvolved and denoised. To go further than the basic denoising approches that don't take care of the correlations between channels, I was told the wavelets can help, but don't know what to do. How would you proceed ?

In each channel (i.e. frequency), the 512x512 image is composed of: a signal strongly correlated between channels and a noise that is stonger than the signal in 90% of the image. The signal is composed of: a point spread function, which is fairly known, convolved with the object, which is different but strongly correlated from a channel to the other.

thank you.