基于粗糙集的图像感兴趣区域自动提取算法  

Automatic extraction of regions of interest based on rough sets

作　　者：孙君顶[1] 陈素霞[1] 郭启强[1] 

机构地区：[1]河南理工大学计算机科学与技术学院,河南焦作454000

出　　处：《传感器与微系统》2009年第3期101-103,107,共4页Transducer and Microsystem Technologies

基　　金：河南省自然科学基金资助项目(2007520019;2008B520012)

摘　　要：在分析常规区域生长算法的基础上,提出一种基于粗糙集理论的感兴趣区域自动提取算法.在显著度图和相对位置指示图的基础上,以像素点的显著度、相对位置及颜色信息构成知识表达系统,通过属性约简和值约简,导出了有用的区域生长决策规则。实验结果表明:该方法与现有算法相比,它能够更好地提取图像感兴趣区域。By the analysis of traditional region growth algorithm, an automatic extraction method of regions of interest（ROI） based on rough sets is presented. Combination of salience and relative position image, knowledge expression system is introduced by the salience, relative position and the color information of each image pixel. The region growth decision rules are educed by attribute and value reduction. The experiment shows that this method can extract the ROI more precisely than the traditional region growth method.

关 键 词：粗糙集理论 感兴趣区域 显著度 提取算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]