| Summary: | 碩士 === 國立中正大學 === 電機工程研究所 === 94 === In this thesis, we propose a method, which combines multiple technologies, including image processing, pattern recognition and expert system, etc. However, we apply it to develop a billiard auto-scoring system. First, we capture the image that players are playing a game by the CCD camera and transmit it to computer. Then, we use image processing technology to analyze and identify the position and image information of each ball. However, it is difficult to analyze group balls by a single image processing technique, so we will analyze and process all kinds of group balls by multiple image processing methods. Our approach includes binary, labeling, edge detection and image identification. Second, we get the number and color of each ball from analyzing image information by pattern recognition technology. There are a lot of differences in the image information because of environment influence, so we propose a training method to overcome the influence. Finally, through the billiards rules organized in the knowledge base in the expert system and the information analyzed by image processing and pattern recognition, the results can be produced by the inference engine. Therefore, we know whether the players follow the rules while playing a game. Moreover, we use the user interface, an audio and visual display of a game, to tell players an analysis result, so that players can easily obtain the status on their games.
|
|---|
