Research On Defect Detection And Classification Method Of Mobile Phone Panel Glass Based On Digital Image Processing

Smartphone cover glass is one of the important components of a smartphone.Global demand for mobile phone cover glass reaches tens of billions each year,and more than 90% of mobile phone cover glass is produced in China.The quality inspection of mobile phone cover glass is very important.Due to the low stability of the existing automated equipment for cover glass detection on the market,the quality inspection of cover glass in the industry still uses manual inspection.In order to improve the stability of the machine vision equipment in the detection of cover glass defects,this paper has studied from cover glass defect image acquisition,defect image segmentation,defect classification,and key defect recognition accuracy improvement.It is mainly reflected in the following three aspects.1.Acquisition of cover glass defect image.This paper first studies the existing image acquisition schemes and analyzes the reasons for their small number of defective image acquisition,unstable image acquisition,and unstable defect identification.In order to increase the types of defective image acquisition,improve the stability of defective image acquisition and the accuracy of key defect discrimination,this paper proposes a scheme for acquiring cover glass defect images based on structured light.This solution acquires two types image sequence of the mobile phone cover glass reflecting structured light—the image sequence reflecting black and white periodic stripes and the image sequence reflecting sinusoidal stripes.The image sequence reflecting black and white periodic stripes has a strong ability to highlight defects.And the enhanced image defect based on the image sequence is stable and rich in variety.Based on the image sequence reflecting sinusoidal stripes,the phase value sensitive to surface damage defects can be analyzed,and the defect information input is strengthened to improve the accuracy of classification of key defects such as scratches and bumps.2.Defective image segmentation.Aiming at the image segmentation of unevenly distributed grayscale of the cover glass enhanced image,this paper first studies the classic image segmentation method in machine vision.and analyzes the uneven distribution of the grayscale of the image.Finally,a defect segmentation method is proposed by mixing the template threshold and the gradient threshold.The template threshold part extracts the overall gentle change trend and local change factors of the defect-free enhanced image as the adaptive threshold for defect threshold segmentation,which overcomes the influence of uneven gray distribution.The gradient threshold part uses the first derivative of the image to highlight the edge information of the defect,recall the low contrast defect,which improves the accuracy of defect segmentation.3.Defect classification.This article first analyzes the defect characteristics of the Smartphone cover glass.According to the different treatment methods for the defects in the production line,the defects are divided into large-area adhesion defects,scratches,linear adhesion defects,bumps,and spots.In order to realize defect classification,the defects are first divided into surface defects,line defects and point defects by using the geometric characteristics of defects.Then,based on the phase analysis,the irregularity information of the defects is introduced,and the linear defects and the point-shaped defects are further divided into scratches,linear attachments,and uneven points and point-shaped attachments.The accuracy of scratch and bump identification is improved by introducing bump information.From the perspective of enhancing the input of defect information,this paper increases the types of defect image acquisition,enhances the stability of defect image acquisition,and improves the accuracy of key defect recognition.This improves the stability of machine vision in the process of detecting cover glass.