摘要
基于人行走时的下肢角度变化包含丰富的个体识别信．幽观点，提出利用下肢角度特征进行步态识别的新方法。对每个步态序列，依据人体解剖学的先验知识定位下肢关节点，计算相邻关节点连线与竖直线的夹角，以此作为下肢角度；通过步态周期分析，提取一个步态周期的下肢角度变化序列作为特征向量表征步态。最后，采用针对小样本问题具有很好分类效果的支持向量机技术实现步态的分类决策。CASIA步态数据库上的仿真结果证明本方法具有较高的识别性能。
Based on the idea that lower-limb angles of motion body contained rich information of human identification, a gait recognition method based on lower-limb angles was proposed in the paper. For each gait sequence, according to the knowledge in body anatomy, the coordinates of lower-limb joints were obtained. Then got four different angles of lower limbs. With analysis of gait cycle, the trajectories of lower-limb angles in one cycle were extracted as feature vectors. Support Vector Machine （SVM） which has an effective classify ability for small sample problem was used for gait classification. Experimental results on CASIA database demonstrate that the approach has encouraging recognition performance.
作者
曾莹
刘波
ZENG Ying, LIU Bo （Eastern Science and Technology College, Hunan Agricultural University, Changsha 410128, China）
出处
《电脑知识与技术》
2010年第01Z期403-405,共3页
Computer Knowledge and Technology
基金
湖南省科技厅科学基金项目（2009CK4010）
湖南农业大学引进人才科学基金项目（08YJ13）
关键词
步态识别
支持向量机
下肢角度
轮廓特征
步态周期
gait recognition
support vector machine（SVM）
lower-limb angle
silhouette feature
gait cycle
