图空间上彩色矢量形态学算子  

Color Vector Morphological Operators on Graph Space

作　　者：李亚宁[1] 王俊平[1] 高艳红[1] 

机构地区：[1]西安电子科技大学通信工程学院,陕西西安710071

出　　处：《电子学报》2015年第3期424-430,共7页Acta Electronica Sinica

基　　金：国家自然科学基金资助项目(No.61173088);西安市产业技术创新计划资助项目(No.CX12485)

摘　　要：针对现有彩色图像形态学矢量排序算法的缺点,将基于图的二值及灰度形态学扩展到彩色图像,在具备完备格的图空间上定义一种新的彩色矢量形态学算子.为使新算子便于实现,结合图论定义最小生成矢量子图函数及相关理论,并在此基础上提出彩色矢量形态学极值提取算法.实验结果表明,该算子在保存彩色信息的完整及关联性方面优于现有的彩色形态学,避免出现"假色"现象及人为决定主导颜色分量的缺点,同时有效地保护了原图像的纹理不发生较大的变化,较好的保存图像的拓扑结构,有较广的应用前景.Since vector sorting algorithms in color image morphology are not mature currently, some new color vector mor- phological operators on graph space possessing of complete lattice are defined, which are based on the binary and grayscale morphol- ogy. The proposed operators can effectively prevent the original image texture from changing and save the topology of the color im- ages. Moreover, combined with graph theory, a minimum spanning vector subgraph function and related theories are defined to im- plement the operators, and the extremum extraction algorithm suitable for color vector morphology of graph space is presented. Ex- perimental results show that the proposed operators have better performance in preservation of color image information integrity and relevance than the existing methods, and that the operators can overcome the shortcomings of false color and human-decision domi- paring color components, which have a broad development trend.

关 键 词：矢量排序 完备格 图论 极值提取算法 拓扑结构 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]