摘要
为了解决当前人体运动识别方法受到复杂背景、可变光照及视角变化的影响,无法准确识别人体运动轨迹的问题,通过特征匹配研究人体运动轨迹识别问题。通过背景提取与差分二级化对人体运动区域进行分割,在此基础上,把人体运动空间描述转换至人体运动关节空间坐标系。通过归一化位移向量序列标识关节活动幅度轨迹,将Fisher向量作为特征,为人体运动轨迹识别提供依据。关节活动幅度轨迹识别选用DTW(Dynamic Time Warping,动态时间归整)方法,获取参考模板与测试模板间的最小累积失真量,将测试模板归类于全部累积失真量最小的一类中,以实现对不同人体运动轨迹长度模板的匹配。结果表明:所提方法识别的人体运动轨迹和实际轨迹基本吻合,受外界环境的影响较小;所提方法与其它方法相比识别率较高,且识别时间较短。可见所提方法识别结果准确,有较强的可行性。
In order to solve the problem that the current human motion recognition method is affected by the complex background,variable light and the change of visual angle,which cannot accurately identify the human motion trajectory,the human motion trajectory recognition is studied by feature matching.Based on the background extraction and differential two-level segmentation of human motion area,the spatial description of human motion is transformed into the spatial coordinate system of human motion joint.The normalized displacement vector sequence is used to identify the trajectory of joint motion amplitude,and the Fisher vector is used as the feature to provide the basis for human motion trajectory recognition.DTW method is used to identify the range of motion of the joint.The minimum accumulated distortion between the reference template and the test template is obtained.The test template is classified into the category with the minimum accumulated distortion,so as to achieve the matching of different human motion track length templates.The results show that the trajectory of human body recognized by the proposed method is basically consistent with the actual trajectory,and is less affected by the external environment.Compared with other methods,the recognition rate of this method is higher,and the recognition time is shorter.It can be seen that the proposed method is accurate and feasible.
作者
于燕山
郭鹏
YU Yanshan;GUO Peng(Sports Department,Xi’an Polytechnic University,Xi’an 710048,China;School of General Education Cangzhou Jiaotang College,Cangzou 061199,China)
出处
《微型电脑应用》
2021年第7期111-115,共5页
Microcomputer Applications
关键词
坐标转换
人体运动
关节活动幅度
轨迹识别
coordinate transformation
human motion
range of motion of joints trajectory
trajectory recognition
