Abstract:

In order to solve the nonconsistence color appearance between different viewpoint images, a variablesize block matching based color correction method for multiview video is proposed. In this paper, firstly the matching blocks are acquired by disparity estimation, and then their reliability is classified according to the residual energy. Secondly, a quadtree segmentation method is adopted to correct and refine the unreliable disparity vectors with variablesize blocks, so the visual artifacts due to incorrect disparity vectors can be eliminated and the object boundaries and structure information can be preserved. Finally, the color correction coefficient is obtained by a linear regression method to correct the color difference between different viewpoints, and histogram equalization is used to further improve the calibration quality. Compared to the histogram matching method and the illumination compensation(IC) method, the experimental results show that the proposed method provides both better subjective and objective quality. It is an effective color correction method for multiview video.

Key words: multiview video;color correction;quadtree segmentation