基于自学习稀疏表示的动态手势识别方法

doi:10.3969/j.issn.1000-436X.2013.06.016

摘要/Abstract

摘要：

针对加速度传感器的手势采集方式提出一种基于自学习稀疏表示的动态手势识别方法。该方法将分类识别问题转化为求解待识别样本对于训练样本的稀疏表示问题，直接对原始加速度信号进行操作，省去了特征提取过程，可方便地添加新的手势类别和删除已有的手势类别；利用面向类别的字典学习，来寻求一个较小的并经过优化的超完备字典来计算待识别样本的稀疏表示，从而大大缩减算法的计算复杂度，满足实时性要求。在包含18种手势的3 000多个样本的公开数据集上进行测试，实验结果验证了该方法的有效性。

关键词: 手势识别, 稀疏表示, 字典学习, 加速度传感器

Abstract:

An approach of robust accelerometer-based hand gesture recognition based on self-learning sparse representa-tion was proposed. This method operated directly on the original acceleration signals by sparse representation without feature extraction and used the class-specific dictionary learning for sparse modeling to reduce the computing cost and time of recognition. The proposed system can easily add a novel gesture category as well as remove existing ones. Ex-periments on real-world database of 18 hand gestures validate the availability of the proposed algorithm.

Key words: hand gesture recognition, sparse representation, dictionary learning, accelerometer

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.

Ling XIAO,Ren-fa LI,Fan-zai ZENG,Wei-lan QU. Gesture recognition approach based on learning sparse representation[J]. Journal on Communications, 2013, 34(6): 128-135.

图/表 8

图1

图2

图3

表1

图4

表2

表3

表4

参考文献 17

[1]	HE Z Y , JIN L W , ZHEN L X , et al. Gesture recognition based on 3D accelerometer for cell phones[A]. IEEE Asia Pacific Conference on Circuits and Systems[C]. Macao, China, 2008. 217-220.
[2]	WU J H , PAN G , ZHANG D Q , et al. Gesture recognition with a 3-D accelerometer[A]. Proceedings of the 6th International Conf on Ubiquitous Intelligence and Computing[C]. Berlin, 2009. 25-38.
[3]	XU R , ZHOU S , LI W J . MEMS accelerometer based non-specific user hand gesture recognition[J]. IEEE Trans on Sensors, 2012,12(5): 1166-1173.
[4]	梁秀波, 张顺, 李启雷等. 运动传感驱动的3D 直观手势交互[J]. 计算机辅助设计与图形学学报 2010,22(3):521-533. LIANG X B , ZHANG S , LI Q L et al. 3D intuitive gesture interaction via motion sensing[J]. Journal of Computer-Aided Design ＆ Computer Graphics, 2010,22(3):521-533.
[5]	王万良, 杨经纬, 蒋一波 . 基于运动传感器的手势识别[J]. 传感技术学报 2011,24(12):1723-1727. WANG W L , YANG J W , JIANG Y B . Motion sensor based gesture recognition[J]. Chinese Journal of Sensors and Actuators, 2011,24(12):1723-1727.
[6]	THOMAS S , BENJAMIN P . Gesture recognition with a Wii controller[A]. Proc of the 2nd International Conference on Tangible and Embedded Interaction[C]. New York, USA, 2008. 11-14.
[7]	TAIWOO P , JINWON L , INSEOK H . E-gesture: a collaborative architecture for energy-efficient gesture recognition with hand-worn sensor and mobile devices[A]. Proceedings of the 9th ACM Conference on Embedded Networked Sensor Systems(SenSys'11)[C]. Ottawa,Canada, 2011. 260-273.
[8]	AKL A , FENG C , VALAEE S , et al. A novel accelerometer-based gesture recognition system[J]. IEEE Trans on Signal Processing, 2011,59(12): 6197-6205.
[9]	LIU J , ZHONG L , Wickramasuriya , et al. uWave: accelerometer-based personalized gesture recognition and its applications[J]. Pervasive and Mobile Computing, 2009,5(6): 657-675.
[10]	MOIZ F , NATOO P , DERAKHSHANI R , et al. A comparative study of classification methods for gesture recognition using a 3-axis accelerometer[A]. Proc of the International Conference on Neural Networks[C]. California, USA, 2011. 2479-2486.
[11]	ARCE F , VALDEZ J . Accelerometer-based hand gesture recognition using artificial neural networks[J]. Trans on Soft Computing for Intelligent Control and Mobile Robotics, 2011,318: 67-77.
[12]	YUAN T , WANG B . Accelerometer-based Chinese traffic police gesture recognition system[J]. Chinese Journal of Electronics, 2010,19(2): 270-274.
[13]	WRIGHT J , YANG Y , GANESH S , et al. Robust face recognition via sparse representation[J]. IEEE Trans on Pattern Analysis and Machine Intelligence, 2009,31(2): 210-227.
[14]	AHARON M , ELAD M , BRUCKSTEIN A M . The K-SVD: a algorithm for designing of over-complete dictionaries for sparse representation[J]. IEEE Trans on Signal Processing, 2006,54(11): 4311-4312.
[15]	WANG L , GU T , TAO X P , et al. A hierarchical approach to real-time activity recognition in body sensor networks[J]. Pervasive and Mobile Computing, 2012,8(1): 115-130.
[16]	LI C G , GUO J , ZHANG H G . Local sparse representation based classification[A]. Proceedings of the 2010 International Conference on Pattern Recognition[C]. Istanbul, Turkey, 2010. 649-652.
[17]	BERG V D , FRIEDLANDER M P . Sparse Optimization with Least-squares Constraints E[R]. Dept of Computer Science, Univ of British Columbia, 2010.

L-SRC	20	K
L-SRC	20	40	60
T=10	99.2	99.3	99.1
T=20	99.2	99.2	99.2
T=40	—	99.1	99.2

分类方法	识别率/%	所用时间/ms
SRC	99.4	280
L-SRC(K=60)	99.3	145
L-SRC(K=40)	99.3	106
L-SRC(K=20)	99.2	70

识别方法	手势种类	识别率/%
识别方法	手势种类	个体相关	个体无关
L-SRC	8～18	100～98.7	99.6～99.3
NN	8～18	99.3～97.7	91.2～88.9
文献[8]	8～18	100～99.8	96.8～94.6
HMM^[6]	8～18	99.9～99.5	75.9～71.5
Uwave^[9]	8	98.6	—