«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

27

期数:

2017年12期

页码:

11-15

栏目:

智能、算法、系统工程

出版日期:

2017-12-10

文章信息/Info

Title:

Research and Realization of Dynamical Gesture Recognition
Algorithm Based on Kinect

文章编号:

1673-629X(2017)12-0011-05

作者:

张莹莹 1 ; 郭 星 2

1. 安徽大学 计算智能与信号处理重点实验室,安徽 合肥 230039;
2. 安徽大学 计算机科学与技术学院,安徽 合肥 230601

Author(s):

ZHANG Ying-ying 1 ; GUO Xing 2

1. Key Laboratory of Intelligent Computing &Signal Processing of MoE,Anhui University,Hefei 230039,China;
2. School of Computer Science and Technology,Anhui University,Hefei 230601,China

关键词:

人机交互; 特征提取; 手势识别; 加权动态时间规整算法;  K 近邻算法

Keywords:

human-machine interaction; feature extraction; gesture recognition; weighted dynamic time warping; K-nearest neighbors

分类号:

TP301.6

文献标志码:

A

摘要:

随着计算机技术和信息化的发展,人机交互在办公以及生活中显得越来越重要。 由于手势具有灵活、直观、简单等优点,成为人机交互研究的重要领域。 针对手势识别技术在自然人机交互中对时间和准确度要求较高的问题,提出一种新的手势识别算法(IDTW-K)。 该算法对经典动态时间规整(Dynamic Time Warping,DTW)算法进行了改进。 利用节点在运动序列中的距离方差对各个节点进行权值动态分配,并对 DTW 的搜索路径进行了详细的分析,采用点和线相结合的
范围约束防止其搜索不合理以及优化 DTW 算法的计算速度, 并结合 KNN 算法提高了手势识别效率。 通过实验对IDTW-K 算法、改进的 DTW 算法和传统的 DTW 算法进行了对比,结果表明所提出的算法在精准度和识别速率上有一定的提高.

Abstract:

With the development of computer technology and informatization,human-machine interaction is becoming more and more im-portant in office and life. Due to the advantages of flexibility,intuition and simplicity for gesture, esture recognition becomes a signifi-cant field in the study of human-machine interaction. According to the problem that gesture recognition has high requirements on time and accuracy in natural human-machine interaction,a new algorithm named IDTW-K is proposed with which the Dynamic Time Warping (DTW) is improved. It makes use of the variance of each node in the action sequence to distribute the weights and analyzes search path of DTW in detail,and then uses the range of constraints in combination of line and point to prevent its unreasonable search and optimize calculation speed of DTW,increasing efficiency of gesture recognition combining K-Nearest eighbor (KNN). The experiments make a comparison of the IDTW-K,the improved DTW and the traditional DTW,which show that the proposed algorithm improves the accuracy and recognition efficiency.

常用功能

更新日期/Last Update: 2018-03-05