基于模糊自适应共振理论的图象分割  

A New Approach for Image Segmentation Based on Fuzzy Adaptive Resonance Theory (FART)

作　　者：黄建军[1] 靳华[1] 赵荣椿[1] 

机构地区：[1]西北工业大学计算机科学与工程系,陕西西安710072

出　　处：《西北工业大学学报》2000年第3期345-348,共4页Journal of Northwestern Polytechnical University

基　　金：国家自然科学基金! (6 9772 0 30 )

摘　　要：提出了一种基于模糊 ART神经网络的灰度门限化图象分割方法 ,该方法不仅可以自动确定分类数目 ,而且还能有效抑制噪声 。Image segmentation is crucial to pattern recognition. In this paper, we for the first time applied a FART neural network to image segmentation. A new gray level thresholding approach to image segmentation based on a fuzzy ART neural network is presented. First we define fuzzy features of the image for segmentation and extract these features. Then we train the FART neural network through these fuzzy features, thus getting the number of classes. Finally each pixel is classed and labeled by the trained neural network to accomplish image segmentation. Experiments prove that this approach can determine automatically the number of classes and suppress effectively the noise in the image(Fig.2). Segmentation result of a real image(Fig.3) demonstrates that our approach is effective.

关 键 词：模糊自适应共振理论 图象分割 模糊神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]