Research And Development Of Vision System For Chess Robot

The development of robotic technology is a mark and externalization of the national high-technology and industrial automation level.Now the application areas of robot rapidly enlarge in production and living. Robots have replaced man in various areas and become more and more important. Robot technology is the hotspot in the world view. The chess robot based on human-computer interaction technology is researched and designed as a typical game robot for the rapid demands of popular science.Chess robot which based on the technologies such as industrial robots and computer visual perception is an interactive system face to extensive mobile staff. This system which based on computer visual perception offers chess automatic identification, and achieves the human-computer interaction in the way of robot which moves pawn automatically. It haves a good application result in popular science fields.chess robot system is still in continuous improvement and development stage currently.The thesis is based on the former reasearcher's work. The principle of chess robot system and some key techniques were deeply researched and analyzed in the thesis.Some Characteristic of image feature and feature extraction and matching which have been used commonly were elaborated in the thesis. The method of chessman recognition based on Moment Invariant and Support Vector Machine is developed on the foundation of researching the location and recognition of chessman which have been used commonly. After that the method based on Moment Invariant and Support Vector Machine is developed and the theory of The Hu moments and The Zernike moments were intensively studied. The characteristic value of writing were extracted used the two kind moments, After that, the results were analyzed and compared. The theory of Support Vector Machine is intensively studied. By the experiment results, it is shown that this method can achieve higher recognition precision in writing recognition. The method of the semi-automatic chessboard location is proposed for the problem of pattern distortion and the algorithm of calculating the coordinates is improved, which reduce the positioning error effectively. In the end, several directions for future researches were provided in this thesis.