Wondering if anyone knows a way (perhaps objective) to identify pixels that are infected with sun glint in an image. I'm using Hedley's 2005 approach in a python to deglint my image but this approach requires selecting pixels over deep waters with glint/specular reflection. I'm having trouble identifying this visually in ENVI - even using different contrast options.

  • One possible approach is to use Mahalanobis distance as described in this paper
    – katahdin
    Feb 22 '17 at 21:09

@user18058 I have thought that you solved your problem about the sun-glint removal. I using Lyzenga algorithm to correct sun-glint in RGBNir images taken using DMC Camera and the results were impressive. If you need help with sun-glint, let me know. Maybe I could help you solve this issue. The first image(left) presents the raw image containing sun-glint. The second image show the glint removal applying the Lyzenga Algorithm for all image and the last image (right) show the results from Lyzenga Algorithm only for the water.

To execute the algorithm only for the water I used a mask based on a NDWI values. It was interesting because the values from pixels out the water were preserving.

enter image description here


Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.