计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (20): 152-158.DOI: 10.3778/j.issn.1002-8331.1706-0285

• 模式识别与人工智能 • 上一篇    下一篇

基于kinect三维骨骼节点的动作识别方法

朱大勇1,2,郭  星1,2,吴建国1,2   

  1. 1.安徽大学 智能嵌入式技术研究中心,合肥 230601
    2.安徽大学 计算机与科学与技术学院,合肥 230601
  • 出版日期:2018-10-15 发布日期:2018-10-19

Action recognition method using kinect 3D skeleton data

ZHU Dayong1,2, GUO Xing1,2, WU Jianguo1,2   

  1. 1.Intelligent Embedded Technology Research Center, Anhui University, Hefei 230601, China
    2.School of Computer Science and Technology, Anhui University, Hefei 230601, China
  • Online:2018-10-15 Published:2018-10-19

摘要: 近年来,基于人体动作识别的应用场景越来越广泛。为了更好的识别效果,提出了一种基于人体三维骨骼节点的动作识别方法。用Kinect等设备获取人体骨骼关节点三维数据信息,以人体臀部为原点重新建立人体坐标系;提取人体关键骨骼的数据信息,定义人体动作特征向量;根据动作表达式用行为树构造动作序列,实现识别。通过对5种定义的动作与其他算法做比较实验,表明提出的方法识别率较高,推广性较强。

关键词: 动作识别, 骨骼数据, 特征向量, 行为树

Abstract: In recent years, the application scene based on human action recognition is more and more widespread. In order to get better recognition effect, this paper presents an action recognition method based on the human body 3D bone node. First of all, it uses Kinect and other devices to obtain the three-dimensional data of the human skeleton, and rebuilds the human coordinate system with the human hip as the origin. Then, it extracts the data of the human skeleton key, and defines the human action characteristic vector. At last, it uses the regular expression to express the human action trajectory, and makes the action sequence with the behavior tree to realize the recognition. The experimental results of 5 kinds of definitions show that the proposed method has high recognition rate and strong generalization.

Key words: action recognition, skeleton data, feature vector, behavior tree