基于共生矩阵的彩色图像增强算法  

Enhancement algorithm for color images based on co-occurrence matrix

作　　者：杨冰清[1] 田小平[2] 吴成茂[2] 

机构地区：[1]西安邮电大学通信与信息工程学院,西安710061 [2]西安邮电大学电子工程学院,西安710061

出　　处：《计算机应用》2012年第9期2573-2575,2584,共4页journal of Computer Applications

基　　金：陕西省自然科学基金资助项目(2009JM8004);陕西省教育厅科技计划项目(2009JK730;2010JK816)

摘　　要：针对传统的直方图均衡化会使图像过分增强,产生不自然的现象,提出一种利用二维共生矩阵的彩色图像增强算法。首先,利用相邻3×3像素之间的相关性来构造共生矩阵;然后,对彩色图像中的亮度分量进行共生直方图均衡;最后,将增强后的图像进行暗原色处理得到比较理想的图像。从所提增强算法与其他3种典型的增强算法的比较可看出:所提算法不仅考虑了图像的全局与局部信息,而且降低了直方图峰值和光环效应,提高了图像对比度,有较好的增强效果。Considering that the traditional method of histogram equalization can make the image produce unnatural enhanced results, a new color image contrast enhancement approach was proposed, dependently using co-occurrence matrix. Firstly, gray correlative characteristics which equalized color components by of pixels were combined in 3 x 3 neighborhoods to construct a co-occurrence matrix. Secondly, the method of co-occurrence histogram equalization was applied to luminance component only and the chrominance components were preserved. Finally, dark channel prior was used to solve the problem mentioned above to obtain the ideal image. Comparing the proposed enhanced algorithm with the other three typical enhancement algorithms, it is shown that the proposed algorithm not only considers the whole and local image information, but also deals with the histogram spikes and halo effect. The experimental results demonstrate that the proposed method can enhance the color images effectively.

关 键 词：图像增强 共生矩阵 暗原色先验 光环效应 清晰度 

分 类 号：TP317.4[自动化与计算机技术—计算机软件与理论]