Author Description

This work concentrates on the edge detection and segmentation methods of texture images, their effectiveness with comparisons. Texture analysis, different texture models, a new and a number of existing texture segmentation methods, edge detection methods (e.g. sobel, canny, susan and statistical methods etc.), clustering techniques (hierarchical and k-means), and feed forward neural networks like self organizing maps have been systematically investigated and explored. A brief overview of these methods and implementation results are given. As a contribution a new statistical texture segmentation technique based on feed forward back propagation neural network has been developed, tested and presented. This new method is believed to be capable of performing finer segmentation on multi texture images, then any other existing techniques.