摘要
提出一种融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矩表达人体轮廓特征,用于描述步态序列的静态特征;依据人体解剖学的知识定位下肢关节点,并提取两脚间的步幅,用于描述步态序列的动态特征;最后,将这两种特征进行组合处理。实验结果表明本文的算法具有不错的识别效能。
A gait recognition algorithm was proposed based on the fusion of static shape feature and dynamic feature.Firstly,modified Hu moments was used to describe body contour as the static feature of gait sequence images.Secondly,according to the knowledge in body anatomy,the coordinates of joints of lower limbs was computed then the stride of feet was obtained as the dynamic feature of gait sequence images.Finally these features were fused.Experimental results demonstrate that the algorithm has efficient recognition performance.
出处
《贵州大学学报（自然科学版）》
2010年第3期81-83,87,共4页
Journal of Guizhou University:Natural Sciences
基金
福建省自然科学基金资助项目(2007J0187)
关键词
步态识别
HU矩
步幅
特征融合
gait recognition
Hu moments
stride
feature fusion
