Study on the Edge Defect Algorithm of Mobile Phone Screen Contour with Self Adaptive Multiple Difference Technology

Yangzhen Ji, Xinmin Ding

College of Nuclear Technology and Automation, Chengdu University of Technology, Chengdu, 610051, China

Abstract: Aiming at the problem that the dust and scratches on the outermost edge of the mobile phone screen are difficult to detect and identify, a global and local dynamic threshold combined multiple image matching differential detection method is designed. Through the use of DLASA linear array camera, we conduct real-time acquisition of high-precision grayscale images, and use the dynamic threshold for image region segmentation and morphological extraction of screen border regions and pixel information points. Then it is contour screening, we conduct multiple differences on the screen’s outermost border area and the overall area of the screen, and the loop is fitted to the judgment in order to finally identify the edge impurities and scratches at the edge of the screen. Combined with the feasibility of the experimental verification algorithm, 300 screens are randomly selected for experimental testing. The detection accuracy reaches 97.3%, the false detection rate is 0.7%, and the leak rate is 1%. The experimental results show that the algorithm can accurately detect the dust and scratches on the outermost edge of the screen of the mobile phone, and provide a reliable solution for industrial screen defect detection, which has certain practical value.

Keywords: Machine vision; Dynamic threshold; Multiple difference; Defect detection