对有计划高协同性多运动员(智能体)团队行为的理解和识别,是计算机视觉研究领域的重要研究问题之一.从视频足球比赛行为识别过程中的目标跟踪与检测、行为表示、分类器的构建及行为识别等4 个方面详细分析和总结了团队行为识别研究国内外现有成果;分析了团队行为识别研究尚待解决的问题及未来研究趋势,为未来的视频足球比赛中团队行为识别及多目标行为识别的研究提供了新思路及文献参考.
Recognition and understanding of team's behavior which is planned and highly cooperative and multi-player (agent) participating is an important field of computer vision research. This paper analyzes and summarizesthe existing domestic and international research achievements in the area of the team's behavior recognition fromfour aspects of target tracking and detection, representation of behavior, construction of classification and behaviorrecognition in the process of the football game. The problems waiting to be solved and some research trends in thefield of team's behavior recognition in the future are also analyzed in this paper. It may provide some novel insightsand literature references for the research of recognition of soccer and multi-objective behavior in the future.