Colour correction of stereo images using local correspondence

A method for enhancing the accuracy of colour correction in stereo images is proposed. The goal is to perform colour compensation at the corresponding points in a stereo image pair using local correspondence information. First, the relevant matching points for colour transfer are extracted using a modified stereo matching algorithm. Then, in order to account for illumination variations and occlusion areas, a new colour transfer method is applied through the weighted sum of the colour difference using local features. The performance of the proposed algorithm has been tested on the reference colour-modified Middlebury dataset.