Abstract: On the basis of the color constancy algorithm based on image derivative framework, a novel color constancy method is proposed based on an improved image derivative framework. The influence of edge type and color channel correlation on illuminant estimation is further considered. a saturation weighting scheme is proposed according to the saturation value of the pixels in the derivative image. An edge taxonomy is put forward based on the photometric properties. The illuminance quasi-invariant is introduced in the algorithm to construct the edge type weighting scheme. It incorporates the weighting schemes into the derivative image frame and accordingly estimates the illumination chromaticity. The original image can be corrected using the von Kries diagonal transformation. Experimental results show that the proposed algorithm can effectively improve the accuracy of the illuminant estimation and the performance of color constancy algorithm.

Key words: color constancy, image derivative framework, edge types, color channel correlation

摘要： 在基于图像导数框架的颜色恒常性算法的基础上，进一步考虑边缘类型和颜色通道间的相关性对于光照色度估计的影响，提出一种基于改进图像导数框架的颜色恒常性计算方法。根据导数图像中像素的饱和度提出一种饱和度权值方案；基于光学特性对导数图像中的边缘进行分类，通过引入照度准不变量提出类型相关的边缘权值方案；将两种权值方案与图像导数框架相结合，并据此进行光照色度估计，利用von Kries对角变换对偏色图像进行校正。实验结果表明该算法有效提高了光照色度估计的准确性，改善了颜色恒常性算法的效果。

关键词: 颜色恒常性, 图像导数框架, 边缘类型, 颜色通道相关性