Defect detection method for TFT-LCD panel based on saliency map model

TFT-LCD can have many defects caused by problems with color filter, TFT array, etc. In this paper, we propose a defect detection algorithm for TFT-LCD panel based on saliency map. Saliency map finds attentional spotlight using three features - color, orientation, intensity. In this work, input images are obtained from the 8bit grayscale line scan camera. Because the input images are grayscale, we use periodicity comes from pixels of TFT-LCD panel and edge operation instead of color information. Some defects of practical defective panel image are too faint to discriminate between normal and abnormal. Experimental results show that our algorithm can distinguish faint defect from poor image.