Research On Panel Micro-defect Detection Method Based On Fusion Of Bright And Dark Fields

Due to the complex background texture,it is difficult to detect small defects with low contrast on the surface of display panels,which brings serious challenges to surface defect detection.In view of the above problems,this project carries out the "research on the panel micro-defect detection method based on the fusion of bright and dark fields".The main research work of this topic is as follows.(1)Theoretical research on multi-angle illumination imaging of bright and dark fields is carried out.Based on the finite-field difference theory,the far-field model of micro-defect scattering under dark-field illumination conditions is established,the influence of the dark-field light source incidence angle on the far-field scattering field energy distribution is analyzed,the theoretical basis for the design of the subsequent bright-and dark-field fusion illumination system is provided,and a bright-and dark-field fusion illumination scheme for micro-defect detection system is proposed.(2)The design of the bright and dark field fusion micro-defect detection system was completed.The bright field tilt illumination is combined with multi-angle dark field illumination,and a high-resolution line array CMOS camera is used for image acquisition,and then a multi-scale transform and pulse neural coupling network based on the system is used for the bright and dark field image fusion method.88.7mm and10?m resolution.(3)A study of low-contrast complex background defect extraction algorithms was carried out.A comparative study of defect enhancement,defect segmentation and boundary extraction algorithms was conducted to analyze the applicability of each method and finally selected the enhancement method based on fuzzy theory,the fuzzy C-mean clustering segmentation method and the Moore neighborhood contour tracking method.System resolution verification experiments and complex texture background and low-contrast surface defect detection experiments were conducted.The experiments show that the proposed method can effectively extract surface defects in complex texture backgrounds such as display panels,and the detection rate can reach 86%.