Abstract: In order to improve the accuracy of defects classification, a texture feature selection method was proposed based on binary partial swarm optimization (BPSO). The first step of this method is collecting and preprocessing defect images, then extracting the texture features to form candidate features. Then the BPSO was applied to select optimal features and redundant features from the candidate features. Finally, the support vector machine (SVM) is trained with these three features to classify defects, respectively. The experiments show that the classification accuracy of optimal features is greatly better than the other two features; demonstrating that the method is feasible and effective for feature selection of fabric defects.