一种基于三维直方图和RFKCN的图像分割方法  

An Image Segmentation Method Based on 3-D Histogram and RFKCN

作　　者：陈子阳[1] 王保平[2] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004 [2]西北工业大学无人机特种技术国防重点实验室,西安710072

出　　处：《计算机学报》2011年第8期1556-1562,共7页Chinese Journal of Computers

基　　金：国家自然科学基金(61040023;61073106;61073060)资助~~

摘　　要：该文提出了一种基于三维直方图和抑制式模糊Kohonen聚类网络(RFKCN)的图像分割方法.该方法分为2步,第1步是像素的模糊化,通过模糊均值和模糊中值构造两幅冗余图像;第2步是通过冗余图像和原始图像组成一个三维特征矢量集,并利用RFKCN聚类网络对该特征矢量集进行聚类,从而达到图像分割的目的.由于新方法不仅利用了图像像元点的灰度分布信息,而且充分考虑了像元点之间的灰度相关信息及其模糊信息,构造出三维特征空间,根据各信息间的竞争性、冗余性和互补性,进行有效的融合,增加了聚类分割的精确度.理论分析和实验表明,该方法具有良好的分割性能.A method of image segmentation based on 3-D histogram and restrain fuzzy kohonen clustering network （RFKCN） is proposed. The method includes two steps. The first step is the fuzzy pixels process in which two redundant images are built by fuzzy mean value and fuzzy medi- an value. The second step is the processing of image segmentation by RFKCN clustering network with original image and their redundant images. The new method not only uses gray distribution information of pixels, but also uses relevant information and fuzzy information of neighboring pixels, to construct 3-dimension observation space. Based on the competition, redundancy, com- plementation and fuzzy character of all information, this algorithm improves the accuracy of clus- tering through efficient fusion. The theoretical analysis and experimental results show that the new algorithm obtains better performance in image segmentation.

关 键 词：图像分割 三维直方图 边缘检测 抑制式模糊Kohonen聚类网络 

分 类 号：TN391[电子电信—物理电子学]