机器人水球比赛中的目标跟踪
内容摘要：
针对原有机器人水球比赛的视频多目标跟踪技术的不足，提出一种基于CamShift算法的带有自适应能力和主动纠错能力的局域目标跟踪方法。通过介绍基本的CamShift算法，并引入自动白平衡调节和自适应能力的优化处理对跟踪方法进行改进，对跟踪过程中可能出现的5种可能错误情形提出具体纠错方案。实验结果表明：该方法有效且算法处理速度能达到实时图像处理的要求。
关键词：
机器人水球比赛；目标跟踪；全局视觉；水下机器人；多机器人协作
中图分类号：TP242.6 文献标志码：A
Abstract: Aiming to the shortage of the existent multiple visional targets tracking technology of the robot water polo game, a local target tracking method is presented based on CamShift algorithm, which is with adaptive ability and active adjustability. Firstly, the basic CamShift algorithm is introduced. Then, the tacking method is improved by including autonomous white balance adjusting and adaptive optimization. Moreover, for the five possible cases which may lead to wrong tracking, correcting mechanism are given in detail. The experiment results show that the obtained method is valid and satisfied for real time image process.
Keywords: robot water polo game; target tracking; global vision; underwater robot; multi-robot cooperation
