基于监控系统的人体行为识别技术研究

针对传统行为识别技术实时性、鲁棒性较差等问题，提出了一种高效鲁棒性的人体行为识别算法。通过基于Meanshift和Kalman滤波相结合的跟踪算法来跟踪定位人体目标；利用肢体特征和区域特征来提取运动特征；利用基于OAA的支持向量机分类识别。仿真实验表明，该算法实时性好、鲁棒性高，能有效应用于监控系统中。

Research of Human Action Recognition Based on Monitoring System

In view of the traditional behavior recognition technology＇s problem of poor real-time performance and robustness, this paper proposes a kind of efficient and robust human behavior recognition algorithm. Meanshift and Kalman is used when the human behavior is tracked real-timely, and selected area of regional and joint angles of limbs to represent human movement, on the target classification step, the OAA-SVM is set up. Simulation experiments show that this method has better robustness and the real-time performance, and can be effectively to the monitoring system.