Texture Segmentation Using Multichannel Filtering And Kohonen’s Self Organizing Map

Abstract

The texture is very important cue in region based segmentation of images. Texture features play a very important role in computer vision and pattern recognition. Texture segmentation can be broken down into two areas, feature extraction and clustering. In this paper, we implement two stage of feature extraction technique using multichannel filter and Self Organizing Map (SOM). Firstly, we go through channel filters, also known as 2-D Gabor functions. The texture features are extracted using a multichannel approach. The channels comprise of a set of Gabor filters having different sizes, orientations, and frequencies to constitute feature vector. This feature vectors are then given to SOM for feature reduction. From the result we can analyze that the disadvantage of using Gabor filters in texture analysis, i.e., the higher dimensionality of the Gaborian feature space is overcome by the reduction in the dimensionality of the feature space achieved by SOM. It is found that the incorporation of secondary feature vectors improves the classification performance and reduces the complexity of the feature extraction and classification stages.  Keywords: texture segmentation, Gabor filters,  frequency