A framework for improving misclassification rate of texture segmentation using ICA and Ant Tree clustering algorithm

Texture segmentation is one of the most challenging problems in the field of image segmentation. Segmenting multi-textured image into different classes of textured region with a minimum rate of misclassification is a challenging issue. This paper proposes a framework for improving misclassification rate by using ICA for designing filter bank and Ant Tree Clustering algorithm, inspired by the self assembly behavior of ants to cluster the feature vectors for texture segmentation. The experimental results shows that misclassification rate of proposed framework is improved to 0.33% using 14 filters as compared with ICA using K-means clustering on Brodatz texture album database.