Research On Large-scale Cabinet Surface Defect Inspection System Based On Machine Vision

The defects on the surface of the product cabinet will not only seriously affect the sales of the product,but also increase the probability of product rework and return,resulting in economic losses.The company is paying more and more attention to the surface quality inspection of the cabinet.However,there is little research on the visual inspection of surface defects for large-size products.This article is based on the specific requirements of an uninterruptible power system(UPS)cabinet manufacturer in Shenzhen for large-scale cabinet surface defect detection systems,based on machine vision.The research on large-scale cabinet surface defect detection system focuses on low-contrast image enhancement,edge detection optimization,and feature point matching.The main tasks are as follows:(1)Research identified the detection specifications of the large-scale cabinet surface defect detection system,analyzed the principle of surface defect detection,analyzed the key factors for obtaining high-quality images,designed a reasonable lighting scheme and the structure of the image acquisition system,and analyzed the type of the light source.The choice of lighting source,light source color and angle of light,reasonable selection of industrial cameras and lenses,and finally built a machine vision based product surface defect detection system.(2)For the problem that low-contrast defects cannot be detected by traditional detection methods,an evaluation method for measuring image contrast is proposed,the principle and parameter significance of Gabor transform are analyzed,commonly used image enhancement methods are studied,and comparative experiments are performed.It shows that the contrast of the low-contrast defect enhanced by the Gabor transform is improved,and the robustness of the defect detection system to light interference is enhanced.(3)The main basis for evaluating the quality of edge detection is clearly defined.The basic principles of typical edge detection algorithms are analyzed and compared.For the noise and discontinuity problems in edge detection,an anchor-based edge is studied.The optimization algorithm was tested and experiments were carried out.The experimental results show that the optimized algorithm can accurately detect the edge of the cabinet image and has high edge quality,which is universal.(4)Researched the image matching technology,deduced the Le image transformation model,analyzed the principle of the SIFT feature point extraction and matching algorithm,studied the RANSAC feature point purifying algorithm,carried out the experiment of extracting and matching the feature points on the surface of the cabinet,and the experiment showed that the purified SIFT The algorithm can effectively extract the feature points of the cabinet surface,and the image matching results are good.Based on the image difference method,the image defect detection after registration can be realized.The research results show that the designed defect detection system can identify the defects such as pits,scratches,and trademarks on the surface of the cabinet.It can detect the fine defects within 1mm of the width of the defects,and verify the feasibility of the system scheme and the effectiveness of the algorithm.With engineering application value.