Automated Fabric Defect Detection Based on NSCT and Twin SVM

A novel detection approach is proposed to detect various uniform and structured fabric defects based on Non-Subsampled Contourlet Transform (NSCT) and Twin Support Vector Machine (TSVM) model. The inspection algorithm is composed of two phases. First of all, the NSCT and the mixture of the generalized Gaussian model are used to obtain compact and accurate signatures for fabric texture description, and TSVM results to learn signatures of defected and non-defected classes. In the second phase, defects are detected on new images via the trained TSVM and an appropriate decomposition of images into blocks. The performance of the proposed detection method is evaluated off-line through extensive experiments based on various types of fabric. Experimental results reveal that the proposed method is effective and robust, and superior to recent state-of-the-art methods in terms of the high detection rate and low false alarm rate.