C.-S. Lin, C.-W. Tsai, C.-M. Tseng, C.-C. Chiu, and S.-C. Chang
thin-film transistor (TFT), liquid crystal display, pattern recognition, gradient orientation code
This paper proposes automatic online inspection and measurements for etching transistors in thin-film transistor liquid crystal display (TFT-LCD) panels. A new method of pattern recognition technology is used to track the target pattern image of TFT etching transistors, to match with the system database, to judge the kind of target etching transistor, and to locate the central position of etching transistors. Furthermore, the control massage is delivered to the servomechanism to normalize the inspection processes and to deal with the measurements for line width and distance. In the process of pattern searching, analysing, and recognizing, due to the characteristics of huge calculations on pattern matching, it is necessary to greatly reduce the time and to find out the correct target pattern image with binary tree structure and binary-tree searching and 16 gradient orientation code algorithms. Template matching method is a simple and powerful tool for pattern recognition and searching. This study also describes the improvement in template matching to greatly reduce matching time without sacrificing accuracy. Additionally, this process could check several items at the same time such as, bias in the centre of illumination, uneven illumination distribution and multiple target detection. After transformation by the gradient orientation code, it can steadily conduct the tracking recognition and positioning of target pattern. The innovation research combines binary-tree searching method and 16 gradient orientation code estimation is more precise. The proposed method can be used for inspecting the defect of LCD glass plates, and the thin film quality can be quickly obtained through automation.
Important Links:
Go Back
