物联网学报 ›› 2019, Vol. 3 ›› Issue (1): 97-105.doi: 10.11959/j.issn.2096-3750.2019.00094

• 服务与应用 • 上一篇    下一篇

智能穿戴设备基于动态模板匹配算法的3D手势识别

李云鹤   

  1. 肇庆学院,广东 肇庆526061
  • 修回日期:2019-03-08 出版日期:2019-03-01 发布日期:2019-04-04
  • 作者简介:李云鹤(1983- ),男,黑龙江齐齐哈尔人,博士,肇庆学院副教授,主要研究方向为物联网、网络编码、稀疏信号表征与处理。
  • 基金资助:
    广东省自然科学基金资助项目(2018A030313346)

3D gesture recognition based on dynamic template matching algorithm for intelligent wearable devices

Yunhe LI   

  1. Zhaoqing University,Zhaoqing 526061,China
  • Revised:2019-03-08 Online:2019-03-01 Published:2019-04-04
  • Supported by:
    The Natural Science Foundation of Guangdong Province(2018A030313346)

摘要:

随着物联网设备的日益普及,智能穿戴设备行业发展迅速,其中以腕带类的智能手环、手表为主。智能穿戴设备具有丰富的传感器和一定的计算能力,通过手势识别作为自身以及面向其他物联网设备的人机交互,具有广泛的用户需求。提出基于动态模板匹配算法的3D手势识别系统,通过智能穿戴设备收集用户的特定手势来判断手势的含义,从而利用更自然的人机交互技术实现对智能设备的控制。使用智能设备的运动传感器读取相应的3D手势数据,结合优化的动态时间规整算法来识别手势;基于移动设备的特征和动态编程,通过斜率来界定曲线路径;同时,通过预存储失真阈值减少模板匹配的计算量和手势识别成本。在手机上进行测试,所提算法与传统算法相比,耗时更少,识别效率和精度更高,可以带来更好的人机交互体验。

关键词: 智能穿戴, 手势识别, 人机交互, 动态模板匹配算法, 动态时间规整

Abstract:

With the popularity of Internet of things equipment,the smart wearing equipment industry develops rapidly,in which wristband smart bracelets and watches are the mainstream.Intelligent wearable devices have abundant sensors and certain computing power.As human-computer interaction for itself and other devices of the Internet of things through gesture recognition,they have a wide range of needs.A 3D gesture recognition system based on dynamic template matching algorithm was proposed.The gesture meaning was judged by collecting user’s specific gesture from hand-held smart devices,and the control of smart devices was realized by using more natural human-computer interaction technology.The motion sensor of intelligent devices was used to read the corresponding 3D gesture data,and the optimized dynamic time warping algorithm was used to recognize gesture.Based on the characteristics of mobile devices and dynamic programming,the curve path was defined by slope.At the same time,the calculation of template matching and the cost of gesture recognition were reduced by pre-storage distortion threshold.The test is carried out on mobile phones,compared with the traditional algorithm,the proposed algorithm takes less time,has higher recognition efficiency and accuracy,and can bring better human-computer interaction experience.

Key words: intelligent wearing, gesture recognition, human-computer interaction, dynamic template matching algorithm, dynamic time warping

中图分类号: 

No Suggested Reading articles found!