Suppose we have a sequence of still images each of which has been contaminated by some particles(ex, dust/sand/smoke) making the images very poor in certain areas.
What approach would be best to teach image regeneration using multiple frames? The simplest technique is to simply find a way to detect what parts of the image are contaminated and uncontaminated and pull uncontaminated sections from each frame.
For spec based contamination you could probably use one of the classic denoising techniques but I'm thinking of having whole sections be contaminated making these single image local approaches not work great.
I am thinking some form of deep learning restoration(advice on architectures would be great), but I am open to more classic signal processing techniques.