基于边界特征的一维最大熵图象分割算法的研究与实现  

Research and Implemention of One-Dimensional Maximum-Entropy Threshold Image Segmentation Based on Edge Features

作　　者：田俊霞[1] 穆国燕[1] 陈树中[1] 

机构地区：[1]华东师范大学计算机系,上海200062

出　　处：《计算机工程与科学》2002年第6期46-47,64,共3页Computer Engineering & Science

摘　　要：本文提出了一种新的图象分割算法 ,该算法首先检测边缘 ,在边界图象的基础上进行图象二值化 ,保留了边界特征 ,而且能自适应地选择阈值 ,克服了一维最大熵阈值方法进行图象分割时丢失边界特征的缺点。大量实验表明该算法取得了很好的效果 ,而且可以处理低质量或边缘模糊的图象 。In this paper,a new method of image segmentation is proposed.The new algorithm detects image edges first.Based on edge images,it processes binary conversions,reserving edge characteristics.And it can implement adaptive threshold selection, overcoming the shortcomings of the one dimensional maximum entropy threshold algorithm that loses the edge characteristics of the image.In addition,this algorithm is efficient in processing low quality images or fuzzy edge images.Many simulation results indicate that high performance is reached and this scheme is practical.

关 键 词：边界特征 图象分割算法 二值化 全局阈值 局部阈值 一维最大熵 边缘检测 图象识别 计算机视觉 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]