基于多实例运动学特征学习的动态手势识别研究

doi:10.11959/j.issn.1000-436x.2017211

通信学报 ›› 2017, Vol. 38 ›› Issue (11): 103-110.doi: 10.11959/j.issn.1000-436x.2017211

基于多实例运动学特征学习的动态手势识别研究

周彩秋¹,杨余旺¹(),庞海波²

¹ 南京理工大学计算机科学与工程学院，江苏南京 210094
² 郑州大学软件与应用科技学院，河南郑州 450002

修回日期:2017-09-21 出版日期:2017-11-01 发布日期:2017-12-13
作者简介:周彩秋（1982-），女，黑龙江哈尔滨人，南京理工大学博士生，主要研究方向为物联网技术及安全、模式识别等。|杨余旺（1966-），男，江苏南京人，博士，南京理工大学教授，主要研究方向为物联网安全、网络编码、大数据等。|庞海波（1979-），男，河南安阳人，博士，郑州大学讲师，主要研究方向为计算机视觉和模式识别。
基金资助:
国家自然科学基金资助项目(61640020);国家自然科学基金资助项目(61402420);河南省高等学校重点科研项目资助计划基金资助项目(17A520014);河南省科技攻关计划项目基金资助项目(172102310496)

Research of dynamic gesture recognition based on multi-instance learning of kinematics features

Cai-qiu ZHOU¹,Yu-wang YANG¹(),Hai-bo PANG²

¹ School of Computer Science and Engineering，Nanjing University of Science and Technology,Nanjing 210094,China
² School of Software and Applied Science and Technology,Zhengzhou University,Zhengzhou 450002,China

Revised:2017-09-21 Online:2017-11-01 Published:2017-12-13
Supported by:
The National Natural Science Foundation of China(61640020);The National Natural Science Foundation of China(61402420);Higher Education Institutions Program of Henan Province(17A520014);Scientific and Technological Research Project in Henan Province(172102310496)

摘要/Abstract

摘要：

在动态手势特征提取和识别方面，利用运动学模式解决动态手势识别问题，在光流场基础上计算出散度模式，旋度模式，对称模式，反对称模式，梯度张量第二、第三主不变模式，应变张量第二、第三主不变模式以及自旋转张量第三主不变模式；进一步提出一种基于多实例学习的方法，将每一个动态手势的所有运动主模式构成一个动态手势词袋，将未知类型动态手势的运动主模式与词袋空间中对应运动主模式进行相似度计算，利用最近邻方法对手势进行识别。实验结果表明：基于多实例运动学主模式学习的动态手势识别方法取得了较高的识别率。

关键词: 手势识别, 运动学特征, 时空轴降维, 多实例学习

Abstract:

Compared to static gestures,dynamic gestures had some new characteristics.The problems of dynamic gestures recognition was spewed by using kinematics mode,such as divergence modes,curl modes,symmetric and ant-symmetric modes,the second and third principal invariant modes of the gradient tensor,the second and third principal invariant modes of the strain tensor and the third principal invariant modes of spin tensor; Further,a framework based on multi-instance learning was proposed,organize all these principle modes for each gesture were organized to a dynamic gestures bag-of-words,and the similarity between the mode of unknown type dynamic gestures and the all bag-of-words were calculated.Then,the nearest neighbor method was used to recognize the dynamic gestures.The experimental results show that the dynamic gestures recognition based on multi-instance kinematics features principal mode learning methods can obtain a higher recognition rate.

Key words: gesture recognition, kinematics features, temporal axis dimension reduction, multi-instance learning

中图分类号:

TP391

周彩秋,杨余旺,庞海波. 基于多实例运动学特征学习的动态手势识别研究[J]. 通信学报, 2017, 38(11): 103-110.

Cai-qiu ZHOU,Yu-wang YANG,Hai-bo PANG. Research of dynamic gesture recognition based on multi-instance learning of kinematics features[J]. Journal on Communications, 2017, 38(11): 103-110.

图/表 12

图1

图2

图3

图4

图5

图6

图7

图8

图9

图10

图11

表1

参考文献 12

[1]	周航 . 基于计算机视觉的手势识别系统研究[D]. 北京:北京交通大学, 2008.
	ZHOU H . Research on gesture recognition system based on computer vision[D]. Beijing:Beijing Jiaotong University, 2008.
[2]	顾立忠 . 基于表观的手势识别及人机交互技术研究[D]. 上海:上海交通大学, 2008.
	GU L Z . Research on surface based hand gesture recognition and human-computer interaction technology[D]. Shanghai:Shanghai Jiaotong University, 2008.
[3]	KHAN R , HANBURY A , THINGER J et al . Color based skin classification[J]. Pattern Recognition Letters, 2012,33(2): 157-163.
[4]	SUN H M . Skin detection for single images using dynamic skin color modeling[J]. Pattern Recognition, 2010,43(4): 1413-1420.
[5]	MOLINA L , ESCUDERO-Vi?olo M , SIGNORIELLO A ,et al. Real-time user independent hand gesture recognition from time-offlight camera video using static and dynamic models[J]. Machine Vision and Applications, 2013,24(1): 187-204.
[6]	FAN L , . A feature-based object tracking method using online template switching and feature adaptation[C]// The 6th International Conference on Image and Graphics. 2011: 707-713.
[7]	LIU Y , ZHOUW , YIN H G ,et al. Tracking based on SURF and superpixel[C]// The 6th International Conference on Image and Graphics. 2011: 714-719.
[8]	WANG W , TUNG C L . Dynamic hand gesture recognition using hierarchical dynamic Bayesian networks through low-level image processing[C]// The 7th International Conference on Machine Learning and Cybernetics. 2008: 3247-3253.
[9]	SGOUROPOULOS K , STERGIOPOULOU E , PAPAMARKOS N . A dynamic gesture and posture recognition[J]. Journal of Intelligent ＆Robotic Systems, 2014,76(2): 283-296.
[10]	DURBIN P A , REIF B . Statistical Theory and Modeling for Turbulent Flows[M]. Second edition.John Wiley ＆ Sons,Ltd.,Chichester. 2011.
[11]	ALI S , SHAH M . Human action recognition in videos using kinematic features and multiple instance learning[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008,32(2): 288-303.
[12]	庞海波, 李占波, 丁友东 . 基于时间序列手势轮廓模型的动态手势识别[J]. 华南理工大学学报(自然科学版), 2015,43(1): 140-146.
	PANG H B , LI Z B , DING Y D . Dynamic hand gesture recognition based on time series gesture contour model[J]. Journal of South China University of Technology (Natural Science Edition), 2015,43(1): 140-146.

方法	准确率
DTW	0.896 3
IDTW	0.923 4
CI	0.914 6
本文方法	0.937 6

基于多实例运动学特征学习的动态手势识别研究

Research of dynamic gesture recognition based on multi-instance learning of kinematics features

在线阅读

PDF下载

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 12

参考文献 12

相关文章 3

Metrics

推荐阅读 0

[1]	肖玲1,2，李仁发1，曾凡仔2，屈卫兰2. 基于自学习稀疏表示的动态手势识别方法[J]. 通信学报, 2013, 34(6): 16-135.
[2]	肖玲,李仁发,曾凡仔,屈卫兰. 基于自学习稀疏表示的动态手势识别方法[J]. 通信学报, 2013, 34(6): 128-135.
[3]	周航,阮秋琦. 基于ROI分割和相干映射的裸手字母手势识别[J]. 通信学报, 2007, 28(5): 94-101.