A Real Time Vision System for Robotic Soccer

Abstract

Our research is based on Small Robot League of RoboCup, where a global vision system is applied to track eleven moving objects. To obtain the precise motion information of one ball and ten robots, two kinds of calibration are really important: the color calibration and the camera calibration. This paper investigates a K-Mean-clustering-based color calibration algorithm to obtain accurate color identification, and Tsai camera model [2] [5] to get the calibration parameters for image distortion compensation. Special techniques are implemented to apply these algorithms to achieve best results. Experimental results have shown the robustness of our approach. 1.