基于尺寸可变块匹配的多视点视频颜色校正方法

A Variable-Size Block Matching Based Color Correction Method for Multi-View Video

针对多视点视频系统中视点间图像颜色差异的问题,本文提出了一种基于尺寸可变块匹配的多视点颜色校正方法。该方法首先通过视差估计得到匹配块;然后根据残差能量对其可靠性进行分类,对不可靠块采用四叉树分割和可变块视差估计进行重新搜索,逐步消除由于匹配块不可靠性带来的方块效应和边缘模糊现象;最后通过线形回归方法得到颜色校正系数,从而校正不同视点间的颜色差异,并且采用亮度直方图均衡进一步提高了校正质量。实验选用多视点测试序列,将本文方法与直方图匹配算法、JMVM中的亮度补偿算法进行比较,结果表明本文方法对校正图像的主观质量和客观质量均有所提高,是一种有效可行的多视点视频颜色校正方法。

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.