Automatic inspection and strategy for surface defects in the PI coating process of TFT-LCD panels

Purpose - The purpose of this paper is to apply an on-line automatic inspection and measurement of surface defect of thin-film transistor liquid-crystal display (TFT-LCD) panels in the polyimide coating process with a modified template matching method and back propagation neural network classification method. Design/methodology/approach - By using the technique of searching, analyzing, and recognizing image processing methods, the target pattern image of TFT-LCD cell defects can be obtained. Findings - With template match and neural network classification in the database of the system, the program judges the kinds of the target defects characteristics, finds out the central position of cell defect, and analyzes cell defects. Research limitations/implications - The recognition speed becomes faster and the system becomes more flexible in comparison to the previous system. The proposed method and strategy, using unsophisticated and economical equipment, is also verified. The proposed method provides highly accurate results with a low-error rate. Practical implications - In terms of sample training, the principles of artificial neural network were used to train the sample detection rate. In sample analysis, character weight was implemented to filter the noise so as to enhance discrimination and reduce detection. Originality/value - The paper describes how pre-inspection image processing was utilized in collaboration with the system to excel the inspection efficiency of present machines as well as for reducing system misjudgment. In addition, the measure for improving cell defect inspection can be applied to production line with multi-defects to inspect and improve six defects simultaneously, which improves the system stability greatly.