Texture-based Fabric Defect Detection Method

Abstract

This paper proposes a detection method based on textile texture features aimed at achieving surface defect detection in textiles. Firstly, an improved gray-level co-occurrence matrix is used to calculate the statistical texture information of the image, including contrast, dissimilarity, homogeneity, energy, correlation, and ASM. Subsequently, a fully connected neural network model is utilized to recognize and classify the derived texture information. Through a comparison with traditional texture recognition methods, experimental results demonstrate that this method achieves better classification performance. The results indicate that the proposed method achieves a recognition accuracy of 92.5% for fabric defects, which is a 6.5% improvement compared to traditional methods. Furthermore, the use of the improved graylevel co-occurrence matrix enhances the contrast features of image boundaries and enables effective unified calculation of texture statistical information in different areas. The application of this method contributes to practical textile inspection.