Design And Implementation Of Mobile Phone Screen Defect Detection System

With the development of technology,mobile phone has become the era of full screens.The quality of mobile phones is largely affected by defects on the surface of screens.Nowadays,manual inspection can no longer meet the needs of industrial production,because it has many disadvantages such as large detection error,poor reliability and low efficiency.And the automatic detection has become more and more important for various industries.The emergence of machine vision inspection technology has solved the problem of detecting the surface defects of mobile phone screens.The image lighting,image noise and blurred image defect segmentation will affect the detection of mobile phone screen surface defects,which yields low detection accuracy.In order to deal with these problems,contrast limited adaptive histogram equalization and two image segmentation algorithms are applied,which include the Otsu method and the Canny operator method.Based on these algorithms,the images are segmented to two images by the two segmentation algorithms,and are merged into one picture with more complete image information.The advantages of this method is that the defect image contains more feature defect information and improves the accuracy of later defect recognition.The support vector machine algorithm for imbalanced data is presented.By adopting different methods to deal with multi-class data and less-class data,the problems of over-fitting of data and loss of important data are solved.The surface defect detection in this paper mainly includes scratch defects,point defects,and dirt defects.Whether the mobile phone screen is qualified depends on the characteristic parameters of these types of defects,and accurately identifying these types of mobile phone screen defects has high value and significance for engineering applications.