奇异点多门限图象分割算法  

A MultiThreshold Segmentation MethodBased on Singularity Analysis

作　　者：安玮[1] 徐晖[1] 李宏[1] 孙仲康[1] 

机构地区：[1]国防科技大学电子工程学院

出　　处：《计算机工程与科学》1998年第3期50-54,66,共6页Computer Engineering & Science

摘　　要：本文提出了一种基于灰度直方图奇异性分析的多门限图象分割算法。由于图象中灰度变化平缓的区域通常和灰度直方图的尖峰相对应，该算法根据尖峰的奇异性确定分割出尖峰的门限。这种算法还能够有效地消除直方图中的毛刺和小起伏对图象分割的不利影响。实验表明该算法获得了很好的效果，具有一定的推广应用价值。In this paper,a new method of image segmentation based on singularity analysis is proposed.Because the regions with gentle changes in image often correspond to the peaks of gray histgraph,the segmentation thresholds can be determined by analyzing the singularity of these peaks.This method can effectively reduce the disturbance of burrs and ripples.Simulation result indicates that high performance is reached and this scheme has great value to practical applications.

关 键 词：图象处理 图象分割 奇异性分析 算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]