A Research Of LCD Defect Detection Algorithm Based On The Structure Of Array Image Acquisition

As the development of the LCD screen production technology and the increasing market demand, more and more high-resolution LCD screens have been produced, which leads the conventional LCD defect detection system based on machine vision to require the expensive and high-quality camera. So, the cost of LCD defect detection system has increased. To solve this problem, a LCD defect detection system based on the structure of array cameras is proposed in this dissertation. The feature of this LCD defect detection system based on the structure of array image acquisition is using the array cameras instead of a single camera, which makes the performance of a single expensive and high-quality camera distribute on many cameras. The cost of LCD defect detection system based on machine vision could be reduced. The main work of this dissertation is as follows:(1) This dissertation set up the LCD screen defect detection based on the structure of array image acquisition experiment platform which includes: OLED back light, polaroid, array cameras and so on. Through this experimental platform, the clear liquid crystal screen defect images are obtained.(2) The image preprocessing of the obtained LCD screen images includes: image gray scale, image noise reduction and extract the target LCD screen in the background.(3) For the LCD screen image after image preprocessing, the basis matrix of the LCD screen image is obtained by fitting pixel energy of the LCD screen image. Then the methods of surface fitting are researched to get the locations of these defects.(4) The binary images of defects are obtained through the between-cluster variance method after studying some threshold segmentation methods. A new defect classification method is proposed in this dissertation in which the external rectangle and ellipse fitting based on invariant moment is combined.80 ?m minimum defect in the 1920×1080 LCD screen can be detected by the LCD defect detection system based on the structure of array cameras. Besides high Resolution, this system also has high accuracy of defects detection and classification which can meet the requirements of LCD screen detection in industrial production.