一种基于主色外观图的彩色图像分割算法

在对三维传统的颜色直方图(CCH)作出了一系列改进的基础上,提出了一种基于主色外观图(DCG)的新的彩色图像分割方法.首先根据改进后的颜色直方图确立彩色图像的近似主色成分,然后再利用CIE-1976色差公式分别计算出其每个像素与主色之间的距离,并据此建立相应图像的颜色距离直方图(CDH),它精确地反映了图像像素与参考色之间的色彩相似度.为了证实CDH在彩色图像分割中的效用性,又通过进一步地扩展得到了CDH集,或可称为主色外观图.实验结果表明,就精确性、鲁棒性和计算的复杂度而言,基于DCG的分割方法能够得到比传统阈值法和聚类法更好的分割效果.

Color image segmentation algorithm based on dominant color graph

This paper proposed a color image segmentation algorithm based on dominant color graph (DCG) on the basis of the improvement of tri dimensional conventional color histogram (CCH).Dominant color components of a color image were first located according to the improved color histogram, followed by using CIE1976 L*a*b* uniform color space to compute the distance of each pixel with respect to the dominant colors. And then it established the corresponding color distance histogram(CDH),which accurately reflected the degree of color similarity of image pixel with respect to the reference color.To demonstrate the efficiency of CDH,it was further extended into a set of CDHs,called DCG,for the purpose of color image segmentation.Experiment results clearly show that the proposed DCG based color image segmentation method can yield superior partition to other traditional thresholding and clustering methods,in terms of accuracy, robustness and computational complexity.