Textile Flaw Detection and Classification by Wavelet Reconstruction and BP Neural Network

This paper presents a method of textile flaw detection and classification based on wavelet reconstruction and BP neural network. The common two types of textile flaws, namely oil stain and hole, can be detected and classified. The method can handle two types of texture fabrics: statistical textures with isotropic patterns and structural textures with oriented patterns. For the extraction of flaw features, histograms of "hole" and "oil stain" are computed as the input of BP neural network. Some samples are selected for testing, the results show that the proposed method can effectively detect defects and classify the types of defect with high recognition correct rate.