Take the 2-minute tour ×
Signal Processing Stack Exchange is a question and answer site for practitioners of the art and science of signal, image and video processing. It's 100% free, no registration required.

In this case, the picture below shows a color fringe detected on its edges.

enter image description here

I am experiencing the same difficulty on a hyperspectral image, so I am looking for a good algorithm correcting chromatic aberration for hyperspectral images. I understand there are several algorithms for RGB images, but it seems that it's difficult to extend any of them for hyperspectral images.

share|improve this question
    
Just a suggestion: You can probably sample few bands and compute the shifts. Then you can interpolate the shifts for bands between the sampled ones. I am not sure if linear or polynomial interpolation would be more appropriate. More details about the capture process, optics and sensor would be helpful. –  Libor Feb 9 '13 at 16:53
add comment

1 Answer

up vote 1 down vote accepted

Having worked with hyperspectral images from experimental setups, I know it's hard to fix these little details.

The best way my group has to fix alignment and chromatic shifts is to use calibration images: Take a picture of something simple for which you know how the result should look like, and use the same correction on every pictures you take hereafter.

A simple example I'm thinking would be to take a picture of a small white dot against a black background. Every misaligned color will create a colored dot at a different place. You then find the mathematical operation necessary to put all these colored dots back together, creating a single white one (it will probably be a translation with a color dependant magnitude).

IF this shift is always the same from picture to picture, you can implement it in your data reduction code. IF not, you will have to take a calibration picture before every picture you take.

If you're lucky, it will be a translation. If you're not, the transformation might also depend on where you are in the picture. In that case, use a grid of white dots.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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