Research On Cloth Defect Detection Algorithm

Food,clothes,housing and transportation are indispensable parts of people’s lives,and clothes are in the first place.With the increasing improvement of people’s living standards,people’s requirements for the quality of clothes are gradually improving.However,the surface of fabric containing defects will affect the quality of fabric.Therefore,fabric defect detection has become an important link to detect the quality of fabric.In China,most of the small and medium-sized manufacturers use the traditional artifical detection method,whose detection speed is slow,and it has low precision,high cost,at the same time,the labor intensity is too large.Therefore,in this paper,we mainly improve the detection algorithm of fabric defect image enhancement and image segmentation,and simulate and verify the improved algorithm.To solve the problem of weak contrast of fabric defect images,a method to enhance the fabric defect images by integrating homomorphic filtering and contrast-limited adaptive histogram equalization is proposed in this paper.Firstly,according to the characteristics of fabric defects appearing in low frequencies,the exponential homomorphic filtering transfer function is proposed,the defect information and background information of fabric defect images are obtained after filtering.Contrast-limited adaptive histogram equalization is applied to the defect information to highlight the detail information,and linear scaling transformation is applied to the high frequency components.Finally,the background information and the defect information are superimposed by the discrete cosine inversion to obtain the enhanced fabric defect image.The simulation result proves that the information entropy value of this defect image enhancement method is larger and the contrast value is smaller,and the enhanced defect image is rich in detail information and high in contrast.For the problem that the defects of fabric defect images are close to the background,this paper adopts the 2D otsu’s fabric defect image segmentation algorithm optimized by the lion swarm algorithm.Firstly,the two-dimensional maximum inter-class variance method is used to process the images with similar background and the object of fabric defects,it can efficiently improve the resolution of foreground and background.Secondly,the local search capability of lion swarm algorithm is used to improve the calculation complexity and long detection time of two-dimensional maximum inter-class variance method to search for the better solution.Finally,the optimal solution obtained from the lion swarm search is used as the near-optimal threshold of the threshold segmentation to segment the defect image.In the process of research,through the use of public textile image database experiments,through the algorithm and MATLAB,Visual Studio and other software combination,to verify the effectiveness of fabric defect image enhancement algorithm and segmentation algorithm.The experimental results show that the image enhancement algorithm combining homomorphic filter and contrast limitation adaptive histogram equalization can effectively solve the problems of low resolution,low contrast and complex noise of fabric defects.Two-dimensional maximum inter-class variance method optimized by lion swarm algorithm can effectively segment the image defects.In the segmentation,the segmentation precision has been improved to a certain extent,the segmentation time has been reduced by nearly half,the segmentation speed increased by 50%,and the segmentation accuracy increased by 2% to 5%...