Algorithms Design for Machine Vision-based Defects Recognition of Mobile Phone Cover Glass

In order to improve the detection efficiency of Mobile Phone Cover Glass (MPCG), we employed machine vision to construct non-destructive detection system and designed defects recognition and detection algorithms. Firstly, the types of defects and inspection requirements were analyzed, combining with processing technologies of glass. Secondly, in view of wide imaging range and large detection area of cover glass, a feature matching algorithm was designed with corner points and straight lines as matching elements, and two regions of interest (ROIs) were extracted by regional division. Finally, defect features were extracted from the image data and classified by a defect extraction algorithm. In addition, we built human-computer interactive software and performed a virtual simulation for defect recognition from mass of cover glass images. The experimental results showed that the non-destructive detection system based on machine vision has a speed of 15 pieces per minute. Compared with the traditional manual light inspection with high subjectivity and uncertainty, this proposed method possesses a higher detection efficiency and accuracy, stronger resistance to environmental interference, which improves the automatic detection level of special products. It may have great feasibility and prospects in the detection field of MPCG.