Abstract: The automatic machine vision inspection way is studied for the Mura defect of TFT-LCD, aiming at the importance of defect inspection and the shortcoming of manual inspection in liquid crystal display (LCD) process. Mura is local lightness variation with low contrast, blurry contour, uneven brightness and textured background. Based on the Semiconductor Equipment and Materials International (SEMI) standard for Mura and the LCD vision inspection platform, the inspection algorithms are researched, including the textured background suppression method using real Gabor filtering, the adjustment of brightness unevenness with homomorphic transform and independent component analysis, the Mura segmentation using active contour model and level set method and the Mura quantification based on SEMI standard. The automatic inspection process is set up by synthesizing the researches proposed above. The inspection experiments show that the proposed method can suppress the textured background, eliminate the unevenness and moire fringe in background, accurately segment defects and carry out quantitative evaluation. The proposed method is applicable to automatic inspection of Mura defect with good robustness and similar to vision characteristic of human eyes. And, for 50 LCD samples, 48 samples are inspected accurately.

Key words: Active contour model, Gabor filtering, Independent component analysis, Level set, Mura defect