Soccer Robot Recognition Algorithms And Decision-making Simulation

The paper firstly present current development and some basic knowledge on computer vision and pattern recognition. Then a new algorithms of vision subsystem applied to MiroSot in Robocup is emphasized on here, which is based on color and morphology, and makes robots need only two colors to distinguish themselves from opponents. This new algorithms is used to solve a problem that there is so many robots in one game in MiroSot in Robocup that robots can be distinguished themselves from others difficultly only by color. Next an arithmetic for 3D angle tracking of human face is proposed. This algorithms is based on image information of HSI format and edge detection, of which the method of extraction of HSI information is improved on. Human face feature, such as eyes and mouth, are found to determine a plane of human face due to three points can fix a plane by the method of tracking robots, which is used to track human face feature. All these will be applied to extraction of 3D angle of human face. At last, a simulation platform, developed with COM, own flexible property and is easy to use. All practical needs can be met and it is introduced here for it helps to debug stratagem of MiroSot in Robocup.