一个足球机器人视频跟踪算法

A Video Tracking Algorithm for Soccer Robots

随着机器人和人工智能技术的发展,在复杂环境中特别是足球机器人的实时比赛的环境进行识别和跟踪已是一个挑战性的课题,提出了一个基于投影的目标跟踪算法,该算法首先把比赛场地分成了若干区域,在这些区域内搜索一个种子点,然后以此点为起点沿垂直或水平方向在目标区域搜索,计算该点的投影值,根据这些投影值识别出目标的一个顶点(或小球的球心),进而识别出目标;最后,提取出7个运动目标的位置和方向,实验结果表明,该算法与其他方法相比有计算量少、图像处理速度快和易于实现等优点。同时,该算法在仿真系统中得到了检验。

With the development of robot and artificial intelligence, the effective recognition and tracking in the complex background becomes a challenging problem, especially in soccer robots game. This paper presents a novel and efficient target tracking algorithm based on a projection method. The projection method has advantages over others. Due to its reduction of calculation work, it makes image processing efficient, and is easy to the implemented. The competition field is divided into several regions from which a seeded point is extracted to be used as a tracer and is traced into a predicting target region. Then the projection of the tracer is calculated along with the vertical （ or horizontal） direction in the target region. Based on the projections, a vertex of the target （ center for ball） can be recognized. Finally, the position or orientation of the 7 moved targets can be determined. The performance of the proposed method is assessed in the video tracking system. Experimental results are also shown in a computer simulated system.