I want to process an image for which I need the method to be robust to lighting conditions. For that the only thing I can think of is normalising intensity. I did this by dividing the image with average intensity and then scaled the values to be between 0-255. This method doesn't seem to work well. What other approach can I follow to make my processing robust to variation in lighting conditions?

  • $\begingroup$ Does the average intensity varies across the image? $\endgroup$ Nov 11, 2017 at 10:49
  • $\begingroup$ Yeah it varies as there my algo works when there are atleast 2 light sources but fails when there are 1 light sources kf same intensity...so the mean changes $\endgroup$ Nov 11, 2017 at 12:13
  • $\begingroup$ the median and alpha trimmed mean are more robust than the mean $\endgroup$
    – user28715
    Nov 11, 2017 at 18:00
  • 1
    $\begingroup$ I should add, more robust in a statistical sense, but maybe not in your approach $\endgroup$
    – user28715
    Nov 11, 2017 at 18:03
  • 1
    $\begingroup$ It eventually depends on what kind of processing you intend to do, and whether the contrast changes locally or not. Contrast-invariant algorithms come in various flavors. Also, mathematical morphology (opening, erosion, top hat...) is a field whose purpose is the design of contrast-invariant algorithms. $\endgroup$
    – sansuiso
    Dec 3, 2017 at 16:15


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.