Research On Modified Algorithm Of Soccer-Robot Vision Subsystem Of MiroSot

In the process of human being's exploration to the nature world, man always hope to find some machine-computing ways to achieve human being's vision function, and to acquire information automatically from outside world. The related research is concentrated on the fields of computer vision, picture engineering. The vision subsystem of robot soccer game is just based on chromatic image processing, and with a great demand for its real-time performance, then it can be an ideal subject for those above rearch fields.Robot soccer game is a multi-agent system development platform which has been put forward in recent years. The robot soccer game includes the artificial intelligent technology which consists of vision, singal fusion, information processing, action decision and so on. Furthermore it includes different control automation technology, wireless communication network technolgoy, electromechanical integration, sociology, histology and so on. So it is called "a technology war in a small playform".Now there are many kinds of robot soccer game, different kinds of robot soccer game qualify the structure design of robot soccer system. The major content of this paper is to reseach MiroSot vision subsystem. Vision subsystem is a unique way throught which the whole system obtains the imformation of the game. Whether the vision subsystem obtains the imformation of the game fastly and accurately is the key of the whole game, it's capability directly affects the victory or defeat of the game. Vision subsystem needs to process real-time track and identification for chromatic game image, and realize robot vision function.The research contents include:(1) Introduce the structure and the principle of MiroSot four subsystem, and emphasize MiroSot vision subsystem.(2) Discuss color-apperceived, compare the features of color modles, and narrate basic principle of digital image processing and image-captured, image enhancement,image Segmentation, image Recongnition.(3) Discuss offline analytic process of MiroSot vision subsystem, and introduce multi-resolution analyse based on wavelet transform and fuzzy c-means cluster algorithm(FCM).(4) Discuss online recognition of MiroSot vision subsystem, and put up dynamic image filtering algorithm, dynamic window-scanning algorithm, image aberrance-revising algorithm, lost target-revising algorithm.Finally it built up an integrated soccer robot vision subsystem based on above theoretics and methods.