I am stitching slices of unrolled fish-eye images of sewer pipes and I find annoying illumination falloff. The falloff looks linear from left to right, and is more noticeable in certain sets then other . I provide 2 subsets, one where the shadow is visible and one where it can be ignored. Although I know the pipe diameter in all cases, I don't know the pipe material, so I cannot calculate the needed light compensation from them. I welcome any suggestions to illumination problem using just the images, and preferably without knowing all images in advance. I picked images with some features to be obvious that these pipes have particularities here and there.
subset 1 (33% of width is overlapping area for blending):
subset 1 results (visible recurrent light falloff):
subset 2 (33% of width is overlapping area for blending)
subset 2 results (no visible light falloff):
