I have to concur, I could in fact be making craters out of LSBs. It's hard to see much, if it's even there.<br /><br />My processing goes in several stages:<br /><br />1. Remove the nearly vertical stipes, by comparing short columns of pixels centered on a target pixel, and adjust the pixel's gain so that it's column has the same average intensity.<br /><br />2. Aquire an intensity value for each pixel, by averaging the intensity of expanding squares of pixels around the target pixel (each square gets equal weight in the average). This averages out a lot of noise, but also keeps detail related to the pixel's local area.<br /><br />3. Each pixel's intensity is then gained by cubing it, and appling a scaler gain to bring it into a range between 0 to 255.<br /><br /><br />I'm still working on it. I have some ideas for step 1, so that I can get a gain and a bias for each pixel, which might do a better job of getting rid of the stripes. I wish the image had the stripes truely vertical, instead of "mostly" vertical. I wish I could get my hands on the "real" original photo.<br /><br />Step 3 is constantly changing, as I try random things. If I manage to get anything better, I'll post that too.