Studying the research paper Multidirectional Scratch Detection and Restoration in Digitized Old Images, E. Ardizzone et al., 2011:

enter image description here

I have several questions regarding this section (5. Restoration),

  1. How can I obtain the average gradient on an area in an image?
  2. How can I obtain a vector that minimizes the sum of absolute differences (SAD) in an image?
  3. Which block-matching approach is being suggested here?
  4. Which factor determines the distance of the blocks from the scratch line?

I could understand the equation (6), but I could not understand how I can implement it practically like in MATLAB or in any programming language.

  • 1
    $\begingroup$ It might be useful to contact the original author of the paper to get a good answer to your questions. It doesn't look like they describe all the details in the paper (eg. their definition of "average gradient" isn't explicitly given, choice of various algorithm parameters such as window sizes isn't given and many symbols are not defined clearly.) $\endgroup$ – Atul Ingle Aug 21 '16 at 19:39
  • $\begingroup$ @AtulIngle, contacted. Mailer daemon failed. $\endgroup$ – user18425 Sep 1 '16 at 11:42

Your Answer

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