I had some scrappy looking images the other day that I needed clean up for a big project. The problem was really one of drift between lines, not signal noise per se, but none the less, I needed to fix it and noise reduction using ImageJ/FIJI was what I had to do the job.
Our go to favorite algorithm (for images 6K x 6K pixels at 4.5nm/pixel), is usually something like:
- ROF-denoise theta=7 (A TV image filter that preserves edges while blurring internal structure)
- CLAHE 1023 and histogram = 511 (Local contrast enhancement to even out and sharpen contrast. Low contrast images always look blurry, even though the focus and resolution are the same – so no penalty for getting this right.)
- Gamma =0.85 (because I wanted to enhance lower intensity information and not blow out the densest elements).
- Unsharp mask, radius=4 and impact = 10%
I also tried smoothing using Gaussian (2pix) and mean (1pix) and median (1pix) filters for the first step.
The results looked like this: … sorry pics coming
Anyone like to share other approaches?