Research And Application Of Cloth Defect Detection Technology Based On Machine Vision

The textiles are still in the stage of manual testing,though textile industry is an important industry in China.The traditional manual detection methods have some problems,such as slow detection speed,high missing detection rate and high false detection rate.In the view,this paper studies the technology of defect detection of monochromatic cloth based on machine vision.In this paper,the finished monochromatic cloth is the main research object.We have selected industrial line array cameras,line array light sources and other hardware to build application platform,developed a set of fabric surface defect detection system suitable for enterprise production.Seven kinds of defects are detected and classified by the system.The main research work of this paper is as follows:1.In view of the influence of noise,uneven illumination and other factors,different preprocessing algorithms are analyzed.This paper chooses median filter to remove noise and Sobel operator to enhance image edge and top hat transformation to eliminate uneven illumination.Fabric defect detection algorithm based on wavelet transform and Laws texture filtering is proposed for the defect is difficult to detect and Laws texture filtering can not make full use of the information on different space and scale layer.The preprocessed image is decomposed by wavelet transform,and the decomposed image is filtered by Laws texture.Then the filtered image is fused by weighted average.Finally,the image is reconstructed by inverse wavelet transform,and the defect region is segmented by mathematical morphologic method.The results show that the average detection accuracy of the proposed algorithm up to 94.7%,which meets the detection requirements.2.In order to solve the problem that the defect is difficult to classify,this paper proposes a BP neural network classification algorithm based on simulated annealing and Particle Swarm Optimization algorithm.The global statistical features of the defects are extracted in the space domain.The local features of the defects are extracted by the Local Binary Patterns(LBP)and the Gray-Level Co-occurrence Matrix(GLCM).All the features are used as the input of the classifier.Then the global search ability of simulated annealing and Particle Swarm Optimization are used to optimize the BP neural network.Finally,the optimized neural network is used to classify defect.The results show that the hybrid feature extraction method can describe defects better,and the average classification accuracy of the optimized classification algorithm for seven kinds of defects is 95.3%.The results meet the classification requirements.3.Matlab algorithm library,C++ language and Qt application framework are used to develop fabric defect detection system.The result shows that the accuracy of the system is 92%to 93%(more than 90%),and the expected result is achieved.