一种基于模糊熵和FKCN的边缘检测方法  

A Method of Edge Detection Based on Fuzzy Entropy and FKCN

作　　者：王保平[1] 刘升虎[1] 张家田[1] 张艳宁[2] 范九伦[3] 

机构地区：[1]西安石油大学电子工程学院 [2]西北工业大学计算机学院,西安710072 [3]西安邮电学院信控系,西安710061

出　　处：《计算机学报》2006年第4期664-669,共6页Chinese Journal of Computers

基　　金：国家自然科学基金(60572133)资助

摘　　要：根据邻域内的灰度分布是有序的、具有方向性以及灰度实质具有结构性三种图像边缘的基本特征,通过引入模糊熵,构造出了一组基于模糊熵的信息测度分量来定量描述图像的这些边缘特征.利用训练样本获取该组分量,并组成一特征向量对模糊Kohonen聚类网络(FKCN)进行训练.用训练好的FKCN聚类网络直接提取其它图像的边缘.该方法无需确定阈值,对弱边界检测较敏感,在特征的选取上充分地考虑了边缘和噪声的本质区别,具有优异的抗噪性能.The three natural characters of image edge are that the gray intensity distribution in neighborhood of image is ordered, directive, and structurized. Through quoting fuzzy entropy, the authors construct three information measures fuzzy entropy-based to describe the natural characters of image edge. Three component vectors for describing three natural characters of image edge are gotten via trained image samples. Firstly, a fuzzy Kohonen clustering network （FKCN） is trained with some feature vector consist of the three component vectors, and then the edge points in a new image are directly extracted by using the trained FKCN. The method does not need any threshold; more sensitive for weak edge detection; has better anti-noise performance since the influence of noise is adequately considered when the feature vector is selected.

关 键 词：边缘检测 模糊熵 模糊KOHONEN聚类网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]