摘要
提出了一种基于步态能量图和分块双向二维主成分分析进行步态特征的算法。首先对图像序列预处理提取运动轮廓,通过分析区域分布直方图检测出运动周期,生成步态能量图描述步态的空间和时间特性,继而使用分块双向二维主成分提取步态特征用以分类,最后在USF步态数据库上测试,并与其它几个算法进行比较。实验结果显示,该方法有更高的识别率和更低的计算复杂度。
A method for gait feature analysis based on gait energy image and Modular-（2D）2PCA is proposed in the paper.The image sequence is pre-processed first to extract the motion profile,the motion cycle is detected by analysing regional distributed histogram,and the gait energy image is formed to describe the temporal and spatial features of the gait.Afterwards,the modular-（2D）2PCA is employed to extract gait feature for classification.Finally,the proposed approach is tested with USF gait database,and is compared with other algorithms.Experimental results demonstrate that the proposed algorithm performs an encouraging recognition rate and less computation complexity.
出处
《计算机应用与软件》
CSCD
2011年第9期232-234,共3页
Computer Applications and Software
