Texture Image Segmentation Using Pulse Coupled Neural Networks

Abstract

Texture, a representation of the spatial relationship of gray levels in an image, is an important characteristic for the automated or semi-automated interpretation of digital images. Many previous analyses have shown how to discriminate texture images, which include gray level co-occurrence matrix (GLCM), Laws' texture energy (LAWS) and Gabor multi-channel filtering (GABOR) etc. We have devised a new method based pulse coupled neural networks (PCNN) to perform texture image segmentation. We propose a segmentation scheme, using PCNN to extract texture features of image and classified by Fuzzy c-Means algorithm (FCM). For demonstration purpose, this paper compares the discrimination ability of two texture analysis methods: pulse coupled neural networks (PCNN) and Gabor multi-channel filtering (GABOR). Experimental results indicate that our method is superior to Gabor multi-channel filtering for a wide range of texture pairs.