结合区域光流特征的时序模板行为识别

系统仿真学报 ›› 2015, Vol. 27 ›› Issue (5): 1146-1151.

结合区域光流特征的时序模板行为识别

王满一, 宋亚玲, 李玉, 张良

中国民航大学天津市智能信号与图像处理重点实验室, 天津 300300

收稿日期:2014-05-03 修回日期:2014-08-04 出版日期:2015-05-08 发布日期:2020-09-01
作者简介:王满一(1988-),女,黑龙江省哈尔滨市,满族,硕士生,研究方向为视频内容分析;宋亚玲(1990-),女,山东省滕州市,硕士生,研究方向为视频内容分析;李玉(1987-),女,山东省菏泽市,硕士生,研究方向为视频内容分析。
基金资助:
国家自然科学基金(61179045,61350009)

Behavior Recognition Combining Regional Optical Flow Features and Temporal Templates

Wang Manyi, Song Yaling, Li Yu, Zhang Liang

Tianjin Key Lab for Advanced Signal Processing, Civil Aviation University of China, Tianjin 300300, China

Received:2014-05-03 Revised:2014-08-04 Online:2015-05-08 Published:2020-09-01

摘要/Abstract

摘要： 提出一种结合时序模板(运动历史图像和运动能量图像)的小波矩特征和计算视频区域光流特征的方法来进行人体行为识别。小波矩具有平移、旋转、尺度不变性并且具备小波的多尺度特性,能够很好地描述运动历史图像和运动能量图像的形状特征,对于简单的动作识别效果较好,但是它不能体现出动作的速度信息。使用Lucas-Kanade光流法计算视频的区域光流能够描述动作的方向和速度特征,弥补了小波矩缺少动作速度信息的不足。结合两种方法描述动作,基于Weizmann数据库所进行的实验测试结果验证了所提方法的有效性。

关键词: 人体行为识别, 时序模板, 小波矩, 光流计算

Abstract: A method was proposed to recognize human behavior, which combined the wavelet moments of temporal templates and the speed feature by optical flow. The wavelet moments are not only rotation, translation and scale invariance, but also have the multi-scale characteristics of wavelet. The shape feature of the motion history image and motion energy image can be described. To simple motions, it's effectively, but it can't present the speed information. Regional optical flow was calculated to describe the direction and amplitude features of motion by Lucas-Kanade algorithm. It could make up shortcomings of wavelet moment without speed information. Being combined two ways of describing the motion and based on the Weizmann Database, experiment results verify the proposed method effectiveness.

Key words: human behavior recognition, temporal templates, wavelet moment, calculation of optical flow

中图分类号:

TP391

王满一, 宋亚玲, 李玉, 张良. 结合区域光流特征的时序模板行为识别[J]. 系统仿真学报, 2015, 27(5): 1146-1151.

Wang Manyi, Song Yaling, Li Yu, Zhang Liang. Behavior Recognition Combining Regional Optical Flow Features and Temporal Templates[J]. Journal of System Simulation, 2015, 27(5): 1146-1151.

参考文献

[1] Bobick A, Davis J.The recognition of human movement using temporal templates[J]. IEEE Trans. On Pattern Analysis and Machine Intelligence (S0162-8838), 2001, 23(3): 257-267.
[2] Lan Chao, Coenen F, Zhang Bai-ling. Driving posture recognition by joint application of motion history image and pyramid histogram of oriented gradients [C]//2013 International Conference on Mechatronics and Semiconductor Materials (S1022-6680), 2014, 846-847: 1102-1105.
[3] Liu Jianyi, Zhang Nanning.Gait history image: A novel temporal template for gait recognition[C]// Proceedings of IEEE International Conference on Multimedia and Expo. USA: IEEE, 2007: 663-666.
[4] 陈实, 高有行. 一种轮廓变化图像小波矩的步态识别[J]. 西安交通大学学报, 2009, 4(1): 90-93.
[5] Tian Yingli, Liu Qingshan, et al. Recognizing expressions from face and body gesture by temporal normalized motion and appearance features[J]. Image and Vision Computing (S2160-7508), 2013, 31(2): 175-185.
[6] Timotius I, Setyawan I. Hand gesture recognition based on motion history images for a simple human-computer interaction system [C]// Proceedings of International Conference on Graphic and Image Processing, Singapore. USA: SPIE, 2012: 87684M-87684M-5.
[7] 吴艳平, 崔宇, 胡士强. 基于运动图像序列的异常行为检测[J]. 计算机应用研究, 2010, 27(7): 2741-2744.
[8] Hiba H, Sreela S.Detection of abnormal behavior in dynamic crowded gatherings[C]// Proceedings of Applied Imagery Pattern Recognition Workshop: Sensing for Control and Augmentation, Washington DC, USA. USA: IEEE, 2013: 1-6.
[9] Lee Jun, Park J, Seo Y.Emergency detection based on motion history image and AdaBoost for an intelligent surveillance system[M]// Information Technology Convergence. The Netherlands: Springer, 2013: 881-888.
[10] Ahad M, Tan J, Kim H.Motion history image: its variants and applications[J]. Machine Vision and Applications (S1432-1769), 2012, 23(2): 255-281.
[11] Ahad M, Tan J, Kim H, et al. Action recognition by employing combined directional motion history and energy images[C]// Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, San Francisco, CA, USA. USA: IEEE, 2010: 73-78.
[12] Gupta R, Jain A, Rana S.A novel method to represent repetitive and overwriting activities in motion history images[C]// Proceedings of International Conference on Communications and Signal Processing, Melmaruvathur, India. USA: IEEE, 2013: 556-560.
[13] Dan M, Toshitaka M, Koji K, et al. Human motion analysis under actual sports game situations: sequential multi-decay motion history image matching[C]// Proceedings of International Conference on Computer Vision Theory and Applications. Portugal: INSTICC, 2013: 229-236.