基于图像模糊熵邻域非一致性的过渡区提取与分割  

Transition Region Extraction and Segmentation Based on Image Fuzzy Entropy Neighborhood Unhomogeneity

作　　者：王彦春[1] 梁德群[1] 王演[1] 

机构地区：[1]大连海事大学信息科学技术学院,辽宁大连116026

出　　处：《电子学报》2008年第12期2445-2449,共5页Acta Electronica Sinica

基　　金：交通部科学技术资金资助项目(No.200136422504)

摘　　要：本文提出了一种基于图像模糊熵邻域非一致性的过渡区直接提取和分割算法.通过求出图像中介于目标和背景之间的过渡区,根据过渡区直方图获得一个最佳分割阈值.构造了描述模糊划分的隶属函数,模糊熵及其邻域非一致性测度.本文算法摆脱了传统图像过渡区提取算法对噪声比较敏感且受剪切值Llow与Lhigh的限制.理论分析和实验结果表明,本文算法具有速度快,抗噪声性能好和稳健性强等特点,优于现有其它基于过渡区提取和分割算法.A transition region extraction and segmentation method based on image fuzzy entropy neighborhood unhomogeneity （NU-TRES） is presented, which is characterized by its speed, ability to deal with noise and robustness. Obtaining a transition region located between the object and background, an optimal segmentation threshold is attained based on the transition region histogram. The membership degree which describes fuzzy division, fuzzy entropy and neighborhood unhomogeneity measurement are created. NU-TRES is not so sensitive to noise as the conventional wansition region extraction methods, depending no more on L_low and L_high.Analysis and experimental results show that this new algorithm （NU-TRES） significantly outperforms the existing transition region extraction and segmentation methods.

关 键 词：模糊熵 邻域非一致性 过渡区 图像提取与分割 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]