Matching OpenCV Exposure Through Images

Question

I was wondering if it is possible to match exposure across multiple images.

For example, let's say you have 5 images that were taken from different angles. 1-3.5 images are taken with the same exposure, while a sharp image has a slightly darker exposure. When I try to combine them into a cylindrical pan, using ( seamFinder with: gc_color, surf detection, MULTI_BAND blending, Wave correctionetc.), results obtained with a large shadow in the middle of the darkness of the image 4.

I also tried using exposureCompensatorwith no luck.

Since I am taking photos in iOS, maybe I can increase the exposure manually when necessary? But this does not seem optimal.

Has anyone else dealt with this problem?

Answer 1

This method is probably redundant (and not just a little), but a modern, modern way of ensuring color consistency between different images is presented in this article by HaCohen et al.

Their algorithm can correct a wide range of errors in image sets. I implemented and tested it on datasets with big errors, and it works very well.

But, again, I believe this is a way to overflow the panorama.

Answer 2

Sunreef provided very good paper, but it seems redundant due to the complexity of the possible implementation.

, , - , . , , . , , , , , ( , ). OpenCV .

OpenCV detail:: ExposureCompensator ( stitiching ) .

• ( : GAIN GAIN_BLOCKS)
• , , ( ) ( ).
• .

, iOS, OpenCV.