改进的人体运动检测方法  

Improved human motion detection method

作　　者：吕金泽 张元[1] 韩燮[1] LYU Jin-ze;ZHANG Yuan;HAN Xie(School of Computer Science and Technology,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学计算机科学与技术学院

出　　处：《计算机工程与设计》2019年第9期2460-2465,共6页Computer Engineering and Design

基　　金：国家科技支撑计划基金项目(2013BAH45F02)

摘　　要：为改善Kinect采集到的用户运动时骨骼节点位置的跳变随机性以及平滑用户动作过程中各骨骼关节的角度,对运动数据的采集和计算算法进行研究,提出改进的中位值平均滤波算法和改进的EWMA算法。改进的中位值平均滤波算法在滤波输出节点位置计算中引入自适应滤波因子和偏移预测量,改进的EWMA算法在角度平滑计算中引入自适应权重动态分配函数。在算法的数据仿真测试中,分别用改进后的两个算法和各自的原算法进行测试和比较分析,实验结果表明,改进算法保证了节点基本静止时滤波和平滑的稳定性,兼顾用户动作变化时节点滤波预测位置的实时性和角度平滑值输出的实时性,优于同类其它算法。To improve the randomness of the jump of the bone node in the movement of the user collected by Kinect,and to smooth the angle of each bone joint during the user’s movement,the collection and calculation algorithm of motion data was studied,an improved median mean filtering algorithm and an improved EWMA algorithm were presented.Among them,the improved median mean filtering algorithm introduced adaptive filtering factor and migration prediction in the calculation of the position of the filter output node.The improved EWMA algorithm introduced adaptive weight dynamic allocation function in the angle smoothing calculation.In the data simulation test of the algorithm,two improved algorithms and their original algorithms were tested and compared respectively.Experimental results show that the improved algorithm not only ensures the stability of filtering and smoothing when the node is basically at rest,but also gives consideration to the real-time ability of predicting the position and the output of the angle smoothing value when the user’s action changes,which are better than that of other algorithms of the same kind.

关 键 词：骨骼节点滤波 角度平滑 自适应滤波因子 偏移预测量 自适应权重动态分配函数 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]