TFT-LCD uneven brightness correction and recognition of MURA area based on EMD method

In automatic visual inspection of TFT-LCD, MURA defect is difficult to recognize due to uneven brightness of the panel. This paper proposed a preprocessing method to eliminate such unevenness using Empirical Mode Decomposition (EMD). Comparing to existing algorithms, such as the Principal Components Analysis (PCA), this method provided a more integral distribution of the unevenness. Then a grey level nonlinear transformation was proposed to eliminate the unevenness of the original image. Besides eliminating the unevenness, results indicated that the EMD method can further give the upheaval features, as white points and texture features, and graded features, as MURA defect, which suggested that it may be possible to extract the MURA defect in an uneven illumination image by the proposed method.