采用纹理图像对兵马俑褪色的复原方法

Recolor method for the fading terracotta warrior using the texture image

针对兵马俑表面褪色纹理的色彩复原问题,在Welsh算法上作了改进,提出一种基于独立成分分析的邻域标准差混合色彩距离颜色迁移匹配算法.在lαβ颜色空间中,快速独立成分分析算法分别对灰色源图像和彩色目标图像做独立成分分析分析,使变换后的两类图像在颜色数据内部呈非高斯分布,各行向量相互独立;然后通过邻域标准差建立混合色彩距离测度函数来比较确定最佳的像素匹配色彩,完成彩色目标图像到灰色源图像的颜色迁移.实验结果表明,该算法可改善Welsh算法在不同颜色在同一亮度点上的颜色迁移问题,能较好地复原兵马俑灰度纹理图像的色彩信息.

For the recolor problem of the terracotta warrior＇s fading texture , we make an improvement on the Welsh algorithm . By using local standard deviation , this paper presents a new mixed color distance of color transfer algorithm based on independent component analysis （ ICA） In lαβ color space , an ICA analysis of the gray source image and color target image are made using the Fast ICA , so that two types of image data have non‐Gaussian distribution and are independent of each row vector in color space . Then , the best matching color is determined by the mixed color distance measurement function founded on local standard deviation . And the color transfer is completed from color target image to gray source image . Experimental results show that the algorithm can improve the color transfer issues of the Welsh algorithm in different colors points of the same brightness . At the same time , color information of the terracotta warrior＇s gray texture can be better recolored .