摘要
步态识别是通过人走路的姿态进行身份识别,研究提出一种组合步态运动中的人体形状静态特征和动态特征的步态识别算法:使用改进的Hu矩和紧致度表达人体轮廓特征,用于描述步态序列的静态特征;提取大腿间的夹角和长宽比,用于描述步态序列的动态特征;并将这两种特征进行组合处理.实验结果表明:本算法的性能较单一特征的步态识别算法有明显的改善.
Gait recognition identifies people by their walking posture. A gait recognition algorithm based on the fusion of static shape feature and dynamic feature is proposed. Firstly, modified Hu moments and compactness are used to describe body contour as the static feature of gait sequence images. Secondly, the angles of thigh and length-width ratio are got as the dynamic feature of gait sequence images. Lastly, these features are fused. Experimental results demonstrate that the performance of proposed algorithm is much better than those based on single feature.
出处
《广西工学院学报》
CAS
2012年第2期60-64,共5页
Journal of Guangxi University of Technology
基金
福建省教育厅省属高校科研专项计划(JK2010049)
福建省教育厅A类科技项目(JA11216)资助
关键词
步态识别
HU矩
紧致度
角度
长宽比
特征组合
gait recognition
Hu moments
compactness
angle
length-width ratio
feature fusion
