针对多视点图像系统中视点图像颜色不一致的问题,提出了基于规正参数调节的多视点图像规正算法.基于像素均值和直方图均值对图像的动态范围进行分割,在参考图像和需规正图像的动态区间内建立映射关系,并扩展到lαβ去相关颜色空间来求取规正系数,最后进行规正操作和全局亮度恢复.与传统的单视点颜色校正算法相比,本文的规正算法在保持图像视觉效果的同时,能将图像规正到与参考图像颜色基本一致的水平,并且校正图像的均方根误差明显小于无校正图像.
In order to solve non-consistence of color appearance in different viewpoint images, a multi-view image color correction algorithm based on adjustment of compensation parameter was proposed. By dynamic range segmentation based on pixel mean and histogram mean, mapping relations between reference images and current viewpoint images were established, and then images were transformed to laβ decorrelated color space to calculate correction parameters for color correction and global brightness restoration. Compared with traditional single viewpoint color correction algorithm, the proposed algorithm can not only maintain original image visual impression, but also correct the image color to be consistent with the reference image. In addition, the root-mean-square errors of corrected images are less than un-corrected images.