An automatic detection method of Mura defects for liquid crystal display

Automatic detection of Mura defects on liquid crystal display (LCD) is an important application of computer vision, an accurate and efficient approach is implemented in this paper. In the stage of image preprocess, in order to reduce the interference of the structurally textural backgrounds of the screen, we use a Gabor filter to process the noise. In the defect detection stage, aiming at the interference of uneven brightness background on Mura defect recognition, based on the principle of mask uniformity, we propose a method for estimating the brightness distribution of the screen to reconstruct the background of the screen, and do background subtraction using the built background image. Defects are detected by gamma correction and threshold segmentation of differential images. The experimental results show that the proposed algorithm greatly shortens the execution time and achieves a good detection effect compared with the background reconstruction algorithm based on regression fitting.