| 基于动态时间规整的人体动作识别方法 |
| |
| 引用本文： | 傅颖,郭晶云.基于动态时间规整的人体动作识别方法[J].电子测量技术,2014(3):69-72. |
| |
| 作者姓名： | 傅颖 郭晶云 |
| |
| 作者单位： | [1]国家知识产权局专利局专利审查协作北京中心北京,100190 [2]天津大学电子信息工程学院,天津300072 |
| |
| 摘 要： | 为了准确识别人体动作,提出了一种基于动态时间规整的人体动作识别方法。首先,该方法通过KINECT摄像头采集人体动作动态骨架序列构建动作模板。然后,通过动态时间规整算法计算被测试样本与单个模板以及模板组合的序列相似度,从而有效利用动作序列所对应动态骨架的序列特性实现人体复杂动作的识别。比较实验表明所提出方法能够准实现多类动作识别,并对于动作速率等外因具有较强鲁棒性。
|
| 关 键 词： | 动态时间规整 动作识别 模板匹配 |
Dynamic time warping-based human action recognition |
| |
| Fu Ying,Guo Jingyun.Dynamic time warping-based human action recognition[J].Electronic Measurement Technology,2014(3):69-72. |
| |
| Authors: | Fu Ying Guo Jingyun |
| |
| Affiliation: | (Patent Examination Cooperation Center of the Patent Office,SIPO,Beijing,Beijing 100190,China;School of Electronic Information Engineering, Tianjin University, Tianjin 300072,China) |
| |
| Abstract: | To recognize human action,this paper proposes a dynamic time warping-based human action recognition method.First,the dynamic skeleton sequence of individual human action is captured by KINECT to construct action template.Then we calculate the similarity between the test sample and individual template/template set.With this method,we can leverage the spatiotemporal dynamics of sequential skeleton to recognize the action category of each sample.The comparison experiments demonstrates that the proposed method can precisely realize the multi-class action recognition and furthermore this method has strong robustness. |
| |
| Keywords: | dynamic time warping action recognition template matching |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
