摘要
提出一种融合步态运动中人体的静态特征和动态特征的步态识别算法:利用背景减除法得到人体轮廓,通过轮廓图像分段距离来表示静态特征;采用步态图像两脚的步幅和步频来表示动态特征;然后对两种特征进行加法融合、最小值融合、最大值融合和Choquet模糊积分融合。实验表明,两种特征融合后的性能优于基于单个特征步态识别算法,融合后识别率提高了5%到10%。
This paper proposes a gait recognition algorithm based on the fusion of static feature and dynamic feature of human body in mo- tion. It uses the background subtraction algorithm to track silhouette of human body, and express static feature by the distance of profile image segmentation; It adopts the stride and cadence of two feet in gait image to express dynamic feature. Then, these two features are performed the fusion methods of addition, minimum value, maximum value and Choquet fuzzy integral. Experimental results demonstrate that the per- formance of the gait recognition algorithm after fusing two features surpasses that of the algorithm based on single feature, the recognition rate after the fusion executed increases 5% to 10%.
出处
《计算机应用与软件》
CSCD
北大核心
2012年第12期123-126,共4页
Computer Applications and Software
基金
湖南省科技计划工业支撑计划重点项目(2010GK2003)
