OLED Display Panel Oriented Defect Detection Algorithm And System Development

OLED (Organic Light Emitting Display) is considered to be the next-generation display panel due to its unique features:low power consumption, wide viewing angle, self-luminous, easy to manufacture on flexible materials. Complicated manufacturing process make various types of defects may exist on OLED panel. These defects have the characteristics of fuzzy boundaries, irregular shape, low contrast with background, that are hidden in the pattern background. The existing detection methods have the problem of low accuracy and recognition rate, and developing a new automated defect detection system becomes a crucial process in the OLED panel manufacturing for the related enterprises. In this paper, we proposed two algorithms to detect the panel defects corresponding to different image resolution and detecting environment respectively. The main research work includes:(1) Investigate and study on the current state of defect detection in the world range.Through the comparison of implementation conditions and effects of various visual detection algorithms, This paper proposed and pointed out the remain problems on this research area constructed a system structure for the OLED screen defect detection.(2) Describe an iterative subtraction method to detect the pixel-level OLED defect. Considering the pixel defects, Image pre-processing are adopted by the methods of median-filtering and image enhancement. Then, the pre-processed image is processed through the subtracting operation by the image pixel template extracted by multiple iterations,which led to the texture background information removed. Based on this result, the paper proposed an improved k-means clustering algorithm; it can quickly identify the image segmentation threshold and recognize the defect effectively.(3) Aiming at macro-level defect detection requirements on the OLED screen, this paper proposed an independent component analysis (ICA) algorithm to detect the defects. Through the processing of ICA analysis, de-mixing matrix construction, and background reconstruction of a defect-free image, the original image with defects are compared with the reconstructed background, and then the defects can be recognised.(4) Based on the algorithms proposed, an OLED-oriented defect detection system is developed with the tool of Iabview8.2and the IMAQ Vision8.0environment.The system is introduced and tested through several OLED image samples. The test result shows that the system developed is validated and most of detects on the OLED image can be detected.