基于3D骨架和MCRF模型的行为识别

doi:10.3969/j.issn.0253-2778.2014.04.005

中国科学技术大学学报 ›› 2014, Vol. 44 ›› Issue (4): 285-291.DOI: 10.3969/j.issn.0253-2778.2014.04.005

• 论著 • 上一篇

基于3D骨架和MCRF模型的行为识别

刘皓，郭立，易波，王冠中

1.中国科学技术大学物理系，安徽合肥 230026；2.中国科学技术大学电子科学与技术系，安徽合肥 230027

收稿日期:2013-09-11 修回日期:2014-01-01 接受日期:2014-01-01 出版日期:2014-01-01 发布日期:2014-01-01
通讯作者: 郭立
作者简介:刘皓，男，1987年生，博士生. 研究方向：视频信息分析与处理. E-mail: lhnows@mail.ustc.edu.cn
基金资助:
国家自然科学基金(61071173)资助.

Human activity recognition based on 3D skeletons and MCRF model

LIU Hao, GUO Li, YI bo, WANG Guanzhong

1.Department of Physics, University of Science and Technology of China, Hefei 230026, China; 2.Department of Electronic Science and Technology, University of Science and Technology of China, Hefei 230027, China

Received:2013-09-11 Revised:2014-01-01 Accepted:2014-01-01 Online:2014-01-01 Published:2014-01-01

摘要/Abstract

摘要： 针对目前行为识别方法的不足，提出一种基于人体3D骨架和多CRF模型(MCRF)的行为识别方法.3D骨架数据量少且保留了行为关键信息的优点，并具有融合多特征和上下文信息的优势.为此，首先基于3D骨架将人体动作划分为全局运动、手臂运动和腿部运动，通过对动作序列进行多类特征提取，形成多类特征集；然后利用CRF模型对每一特征集建模，再融合所有的CRF模型，得到MCRF模型；最后利用MCRF模型进行行为识别.实验结果表明，该方法具有较高检测率.

关键词: 行为识别, 3D骨架, MCRF, 特征提取

Abstract: Considering the disadvantages of the traditional human activity recognition system, a human activity recognition system using an MCRF model and 3D skeletons was proposed. Its 3D skeleton data has less data and retains the key information, and the MCRF model has the advantage of being able to combine more features and utilizing adaptive contextual information. First, human activity was divided into global activity, arm activity, and leg activity. Several feature subsets were formed through more feature extraction. Then, CRF models were used on each feature subset to generate CRF units. Finally, all the CRF units were combined to produce the MCRF model which was utilized to recognize human activity. The experimental results indicate that the proposed method can improve detection accuracy.

Key words: human activity recognition, 3D skeleton, MCRF, feature extraction

中图分类号:

TP391.4

刘皓，郭立，易波，王冠中. 基于3D骨架和MCRF模型的行为识别[J]. 中国科学技术大学学报, 2014, 44(4): 285-291.

LIU Hao, GUO Li, YI bo, WANG Guanzhong. Human activity recognition based on 3D skeletons and MCRF model[J]. Journal of University of Science and Technology of China, 2014, 44(4): 285-291.

参考文献

［1］ Zhao L, Guo L, Xie J S, et al. Video abnormal target description based on CRF model［C］// International Conference on Audio, Language and Image Processing. Shanghai, China: IEEE Press, 2012: 519-524.
［2］ Wang Y, Mori G. Learning a discriminative hidden part model for human action recognition［J］. Advances in Neural Information Processing Systems, 2008, 21: 1 721-1 728.
［3］ Gu Junxia, Ding Xiaoqing, Wang Shengjin. A survey of activity analysis algorithms［J］. Journal of Image and Graphics, 2009, 14(3): 377-387.
谷军霞, 丁晓青，王生进. 行为分析算法综述［J］. 中国图形图像学报, 2009, 14(3): 377-387.
［4］ Gu Junxia, Ding Xiaoqing, Wang Shengjin. Human 3D model-based 2D action recognition［J］. Acta Automatica Sinica, 2010, 36(1): 46-53.
谷军霞, 丁晓青，王生进. 基于人体行为3D模型的2D行为识别［J］. 自动化学报, 2010, 36(1): 46-53.
［5］ Polana R, Nelson R. Low level recognition of human motion (or how to get your man without finding his body parts)［C］// Proceedings of the IEEE Workshop on Motion of Non-Rigid and Articulated Objects. Austin, USA: IEEE Press, 1994: 77-82.
［6］ Davis J W, Bobick A F. The representation and recognition of human movement using temporal templates［C］// IEEE Conference on Computer Vision and Pattern Recognition. San Juan, USA: IEEE Press, 1997: 928-934.
［7］ Yamato J, Ohya J, Ishii K. Recognizing human action in time-sequential images using hidden Markov model［C］// IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Champaign, USA: IEEE Press, 1992: 379-385.
［8］ Phillips S J, Anderson R P, Schapire R E. Maximum entropy modeling of species geographic distributions［J］. Ecological modelling, 2006, 190(3-4): 231-259.
［9］ Lafferty J, McCallum A, Pereira F C N. Conditional random fields: Probabilistic models for segmenting and labeling sequence data［C］// Proceedings of the 18th International Conference on Machine Learning. Williamstown, USA: Morgan Kaufmann Publisher, 2001: 282-289.
［10］ Zhong P, Wang R S. A multiple conditional random fields ensemble model for urban area detection in remote sensing optical images［J］. IEEE Transactions on Geoscience and Remote Sensing, 2007,45(12): 3 978-3 988.
［11］ Smisek J, Jancosek M, Pajdla T. 3D with Kinect［C］// Consumer Depth Cameras for Computer Vision Advances in Computer Vision and Pattern Recognition. Barcelona, Spain: IEEE Press, 2013: 3-25.
［12］谷军霞. 行为表征与行为识别方法研究［D］. 清华大学，2010.
［13］钟平. 面向图像标记的随机场模型研究［D］. 国防科学技术大学, 2008.
［14］ Shotton J, Winn J, Rother C, et al. TextonBoost for image understanding: Multi-class object recognition and segmentation by jointly modeling texture, layout, and context［J］. International Journal of Computer Vision, 2009, 81(1): 2-23.
［15］ Shotton J, Winn J, Rother C, et al. TextonBoost: Joint appearance, shape and context modeling for multi-class object recognition and segmentation［J］. Lecture Notes in Computer Science, 2006, 3951: 1-15.
［16］ Ramos F, Fox D, Durrant-Whyte, H F. CRF-Matching:condtional random fields for feature-based scan matching［A］// Robotics: Science and Systems, MIT Press, 2007.
［17］ Niebles J C, Wang H C, Li F F. Unsupervised learning of human action categories using spatial-temporal words［J］. International Journal of Computer Vision, 2008, 79(3): 299-318.
［18］ Uddin M Z, Thang N D, Kim J T, et al. Human activity recognition using body joint-angle features and hidden Markov model［J］. ETRI Journal, 2011, 33(4): 569-579.

[1]	马玉莲, 崔文泉. 一种预测高频价格的端到端双目标多任务方法[J]. 中国科学技术大学学报, 2021, 51(3): 246-258.
[2]	张会敏，杨明，吕静. 基于自适应核联合稀疏表示的多特征高光谱图像分类[J]. 中国科学技术大学学报, 2018, 48(4): 298-306.
[3]	饶齐，杨燕*，滕飞. 基于多视图加权聚类集成的高速列车工况识别[J]. 中国科学技术大学学报, 2018, 48(1): 35-41.
[4]	孙建，王鹏，陈宗海. 基于样本分布信息的自适应Hough变换方法[J]. 中国科学技术大学学报, 2015, 45(1): 48-55.

基于3D骨架和MCRF模型的行为识别

Human activity recognition based on 3D skeletons and MCRF model

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 4

编辑推荐

Metrics

本文评价