分层色彩校正算法研究

Research on hierarchical colour correction method

提出了一种基于视网膜皮层理论和颜色视觉理论分层色彩校正算法:引入颜色视觉理论中的三色学说进行全局分类,使用广义高斯混合模型计算全局系数;简化分层色彩校正模型减少计算量;引用retinex理论对三通道分别进行处理,进行高光区域提取;使用对立学说进行色差计算,根据色度距离和空间距离设置系数权重,并根据系数校正像素;采用分层色彩校正模型整合图像。所提算法融合了颜色视觉理论和视网膜皮层理论,对现有的分层色彩校正进行了进一步的改进。实验验证该算法在模拟人类视觉系统色彩恒常性方面具有很好的合理性和实用性,实验表明该算法对非均匀多光源色偏图像有很好的校正效果。

This paper proposes a hierarchical color correction algorithm which is based on the retinex theory and color vision theories. In order to ensure the accuracy of color correction, it uses hierarchical color correction algorithm. To reduce the amount of computation, it simplifies the hierarchical color correction model. In order to get more reasonable overall classification, it introduces the trichromatic theory. And to extract better high-light regions, it refers to the retinex theory which processes three channels separately. While in order to get better pixel correction coefficient, it bases the theory on opponent color theory to calculate chromatic aberration as well as takes the different effects between spatial distance and Euclidean distance into account to calculate the coefficient. In a word, it combines the retinex theory and the color vision theories, and it further improves the existing hierarchical color correction algorithm. What＇s more, the method has been tested on color cast images to testify its rationality and practicality on imitating human being＇s color constancy capacity,and the results suggest that the algorithm works well on non-uniform multi-light color cast images.