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.

  • 1
    $\begingroup$ Are the different frames aligned? $\endgroup$ – Royi Oct 21 '20 at 6:33
  • $\begingroup$ Yes, I mean ideally they could be offset a little but as a starting point they can be aligned. $\endgroup$ – FourierFlux Oct 21 '20 at 6:33
  • 1
    $\begingroup$ In case the images are aligned and you have for each pixel at least majority of frames which are good you can create an array of them and apply median along the concatenation dimension. $\endgroup$ – Royi Oct 21 '20 at 17:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.