河北水利电力学院学报

河北水利电力学院学报 ›› 2023, Vol. 33 ›› Issue (1): 12-18.DOI: 10.16046/j.cnki.issn2096-5680.2023.01.003

• 人工智能与机器人专题 • 上一篇    下一篇

基于Kinect V2的肢体动作识别方法设计与实现

王政博1,2, 唐勇1,2, 刘辰淼1, 孙东来1,2   

  1. 1.河北水利电力学院电气自动化系,河北省沧州市黄河西路49号 061001;
    2.河北省高校水利自动化与信息化应用技术研发中心,河北省沧州市黄河西路49号 061001
  • 收稿日期:2022-08-11 修回日期:2022-08-28 发布日期:2023-10-26
  • 作者简介:王政博(1994-),男,河北沧州人,硕士,研究方向为机器视觉。E-mail:yunworthy@163.com
  • 基金资助:
    沧州市科技计划自筹经费项目(204102006);河北水利电力学院基本科研业务费研究项目-青年科研创新项目(SYKY2019);国家级大学生创新创业训练计划项目(202110085010);河北省高校水利自动化与信息化应用技术研发中心项目。

Design and Implementation of Body Motion Recognition Method Based on Kinect V2

WANG Zheng-bo1,2, TANG Yong1,2, LIU Chen-miao1, SUN Dong-lai1,2   

  1. 1. Department of Electrical Automation, Hebei University of Water Resources and Electric Engineering, 061001, Cangzhou, Hebei, China;
    2. Water Resources Automation and Informatization Application Technology Research and Deuelopment Center of Hebei College, 061001, Cangzhou, Hebei, China
  • Received:2022-08-11 Revised:2022-08-28 Published:2023-10-26

PDF

摘要： 在计算机视觉、模式识别等领域,肢体动作识别一直是一个热门课题。肢体动作识别技术已被广泛地应用于各个行业。本文采用Kinect设备,采集人体20个关键部位的三维坐标,并将其应用于肢体动作的分析和辨识;在此基础上,分析了基于人体骨骼的三维深度信息,并进一步深入地分析了人体运动特征的辨识与表达。通过系统仿真,分析识别结果,得到识别准确率均在92%以上,同时对不同形态的人物进行了识别分析,均可正常识别,验证了系统的可行性。

关键词: Kinect, 肢体识别, 数据采集, 骨骼点

Abstract: In the fields of computer vision and pattern recognition, body motion recognition has always been a hot topic. Body motion recognition technology has been widely used in various industries. In this paper, Kinect equipment is used to collect the three-dimensional coordinates of 20 key parts of the human body and apply them to the analysis and identification of body movements; on this basis, the three-dimensional depth information based on human bones is analyzed, and the human motion is further analyzed. Identification and expression of features. Through the system simulation and analysis of the recognition results, the recognition accuracy rates are all above 92%. At the same time, the recognition and analysis of different forms of characters can be recognized normally, which verifies the feasibility of the system.

Key words: Kinect, limb recognition, data collection, skeleton points

中图分类号: