Research On Online Detection System Of Mobile Phone Cover Glass Defects Based On Machine Vision

Smart phones are becoming more and more important in people’s life.As one of the necessary accessories,mobile phone cover plate also has a large market demand.Manufacturers rely on strict control of product quality to enhance their market competitiveness.Although the production of cover glass to printing has been automated,the quality inspection process still uses manual visual inspection,which cannot accurately quantify defects,and vulnerable to human subjective emotions,low efficiency,high cost and high rate of missed and false detection.In this paper,the characteristics of machine vision,such as high speed,non-contact,high precision,quantifiable and 24-hour continuous work,are used to detect the defects of mobile phone cover glass,which provides guidance and optimization reference for the processing technology of the cover plate,and improve the quality and efficiency of processing.In this paper,the overall design and test scheme are defined after specifying the detection task.The main research work in the full paper is as follows:(1)The hardware module of the machine vision inspection system is designed and built to improve the imaging quality of the inspection system as an entry point to obtain high-quality images and reduce the rate of missed and false detection from the source of defect detection.Based on the imaging characteristics of different light sources and illumination modes,an imaging system with a linear array camera and three light sources is designed.Three images with the same pose can be obtained at one time without image registration.In order to adapt to the actual production line,the design scheme of the conveying platform of synchronous wheel and synchronous belt is proposed.The images are taken at the gap between the synchronous belts to avoid background occlusion.Extract the ROI region of the cover plate for preprocessing the image,and remove the redundant area in the original image.(2)The edge defects are detected by using the characteristics of clear outline of backlight image.The detection effect and efficiency of common edge detection algorithms are compared,and an edge defect detection algorithm based on ROI and sub-pixel is proposed.Firstly,the rough contour is obtained by binarization,and then the ROI region of the edge is obtained to reduce the computational complexity of the algorithm.Secondly,combined with the characteristics of the image,the sub-pixel detection of ROI region based on Roberts operator is carried out to accurately locate the edge.Finally,the edge fitting is compared to identify defects.The software can automatically fit the standard edge without making template images,and can detect various kinds of cover plates adaptively.(3)Aiming at the surface defects of the cover plate,a surface defect detection algorithm based on gray mean and information fusion is designed.Firstly,the image and the gray mean image are different,so that the multi angle light image and the coaxial light image have the same gray characteristics,that is,the background is dark and the defect is bright.Then,the imaging ability of the two images is different,The information is added and fused to enhance the image,which highlights the difference between the defect and the background,and reduces the missed detection rate of the system.Through the experimental evaluation,the online detection system of mobile phone cover glass defects takes 1.711 s on average,and the accuracy rate is 91%,which basically meets the requirements of the detection task.It has a certain practical application value in the field of industrial inspection,and provides a strong support for further research in the follow-up.