Research On Object Tracking Algorithm For Football Robots Based On CamShift Algorithm

Vision system is an important way for football-robots to obtain information, as the basic supporting system, it provides location information of the targets for football-robots, real-time location information helps robots better decide their strategy and direction. This dissertation introduces most of the concepts of comuputer vision field such as image acquiring, color image preprocessing, color image dividing, object identifying and tracking.First we introduce some common methods for object detection, such as temporal difference, background subtraction and algorithms based on background model, then we come to Hough circle shape objects detection, also we introduce the MeanShift and CamShift algorithm.The existing CamShift algorithm can detect and track specified color objects continously, but when applied on football-robots directly, problems such as low efficiency in matching and mismatch arise. Based on the football-robot made by Grandar Co.Ltd, we improved the CamShift algorithm. After adding target location prediction, contour-matching, robots motion-control and enlarging searching window, the new algorithm can overcome enviroment color noise and has better robustness.