Pixel Classification by Divergence-Based Integration of Multiple Texture Methods and Its Application to Fabric Defect Detection

Abstract

This paper presents and evaluates a pixel-based texture classifier that integrates multiple texture feature extraction methods through a new scheme based on the Kullback J-divergence. Experimental results show that the proposed technique yields qualitatively better image segmentations than well-known both supervised and unsupervised texture classifiers based on specific families of texture methods. A practical application to fabric defect detection is presented.