基于权重的形态学腐蚀改进算法  

An Improved Morphological Erosion Algorithm Based on Weight

作　　者：章普生[1] 刘建平[1] 刘国良[1] 

机构地区：[1]国防科学技术大学机电工程与自动化学院,湖南长沙410073

出　　处：《计算机仿真》2008年第6期331-334,共4页Computer Simulation

摘　　要：胰腺癌辅助诊断系统中,采用极限腐蚀法从粘连细胞中得到种子点,并以此为基础构建分离线分割出单个细胞图像。但传统方法中,小半径结构元素的近似表达会造成极限腐蚀过程中误差的累计,最终得到错误的种子点;为解决这一问题,文中提出了基于权重的结构元素构造方法,依据各像素对维持元素形状的作用来设定结构元素,得到形态学腐蚀运算的改进算法。通过Matlab实验,表明新方法能有效减少极限腐蚀的误差累积,取得粘连细胞正确的种子点。In computer- aided medical diagnosis system for pancreatic cancer, in order to separate individual cell from overlapping cells or clustered cells, the ultra - erosion may be used to obtain the seeds, and these seeds will be treated as the probability - of- centre, Then the segmentation will be obtained based on these centres. But in normal morphology, small radius structuring elements can not be accurately expressed, and it will lead to serious error accumulation, so the number or position of seeds may be incorrect. To deal with the problem, this paper proposes an improved morphological erosion algorithm based on weight. The improved algorithm uses different methods from traditional morphology to get new kinds of structuring elements, Compared with the traditional elements and erosion algorithm, the experiment using Matlab shows that the improved algorithm can reduce error accumulation, and get the right seeds.

关 键 词：计算机辅助诊断 细胞分割 数学形态学 结构元素 极限腐蚀 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]