T.-K. Lin (Taiwan)
Pattern recognition, moment invariants, light-guide plate, and LCD
Liquid crystal display (LCD) has become a dominant display device. Defects of dot patterns on light-guide plate of LCD can cause visual failure. In this work, a method for automated inspection of LCD light-guide plate is presented. Dot-pattern images of the light-guide plate are preprocessed using image segmentation. Image recognition is then performed using a moment-based approach. Through this approach, parameters for the recognition are proposed to reduce time expense and storage space for the inspection. It is found that different dot patterns of LCD light-guide plate can be successfully classified using this method. The parameters such as image scale and number of shape descriptors can be greatly reduced. By the moment-based approach, low data demand for the inspection of different defects on the plate can be achieved.
Important Links:
Go Back
