OLED Defect Inspection System Development through Independent Component Analysis

Organic Light Emitting Displays (OLED) is a new type of display device which has become increasingly attractive and popular. Due to the complex manufacturing process, various defects may exist on the OLED panel affecting the quality and life of the display panels. These defects have the characteristics of fuzzy boundaries, irregular in shape, low contrast with background and especially, they are mixed with the pixel texture background increasing the difficulty of a rapid recognition. In this paper, we proposed an approach to detect the defects based on the model of independent component analysis (ICA). The ICA model is applied to a perfect OLED image to estimate its corresponding independent components (ICs) and create the de-mixing matrix. Through estimation and determination of a proper ICi row vector of the faultless image, a new de-mixing matrix can be generated which constitutes only uniform information and is then applied to reconstruct the texture background of source OLED images. Through the operation of subtraction of the reconstructed background from the source images and the binary segmentation, the defects can be detected. Based on the algorithms, a defect detection system for the OLED panels was implemented and the testing work was performed in this study. The testing results show that the proposed method is feasible and effective.