[1]李拟珺,程旭,郭海燕,等.基于多特征融合和分层反向传播增强算法的人体动作识别[J].东南大学学报(自然科学版),2014,44(3):493-498.[doi:10.3969/j.issn.1001-0505.2014.03.008]
 Li Nijun,Cheng Xu,Guo Haiyan,et al.Human action recognition based on multi-feature fusion and hierarchical BP-AdaBoost algorithm[J].Journal of Southeast University (Natural Science Edition),2014,44(3):493-498.[doi:10.3969/j.issn.1001-0505.2014.03.008]
点击复制

基于多特征融合和分层反向传播增强算法的人体动作识别()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
44
期数:
2014年第3期
页码:
493-498
栏目:
计算机科学与工程
出版日期:
2014-05-16

文章信息/Info

Title:
Human action recognition based on multi-feature fusion and hierarchical BP-AdaBoost algorithm
作者:
李拟珺程旭郭海燕吴镇扬
东南大学信息科学与工程学院, 南京 210096
Author(s):
Li Nijun Cheng Xu Guo Haiyan Wu Zhenyang
School of Information Science and Engineering, Southeast University, Nanjing 210096, China
关键词:
特征提取 动作识别 反向传播增强算法 神经网络 分层识别
Keywords:
feature extraction action recognition back-propagation(BP)-AdaBoost algorithm neural network hierarchical recognition
分类号:
TP391.4
DOI:
10.3969/j.issn.1001-0505.2014.03.008
摘要:
为了推广神经网络在人体动作识别中的应用,设计了一种基于分层识别框架和增强算法的动作识别系统,该系统融合了光流直方图、有向梯度直方图、Hu的矩特征、分块剪影和自相似矩阵等多种特征.为了与反向传播网络的增强相匹配,将传统的二分类增强算法扩展到多分类版本.此外,系统采用了包含预判决和后判决的分层识别框架,前者通过分析运动显著区域的位置,把动作粗分为几个子类,后者则利用额外的特征进一步提高识别准确率.基于Weizmann和KTH数据库的实验结果表明:神经网络相对于常用的支持向量机具有明显的优越性;结合分层识别的反向传播增强算法可以极大减少运算代价与动作类间的混淆,识别准确率较高.
Abstract:
To popularize the application of neural network in human action recognition, an action recognition system based on the hierarchical recognition framework and the boosting algorithm is designed, which mixes together multiple features such as histograms of optical flow, histograms of oriented gradients, Hu’s moments, block-silhouettes and self-similarity matrices. To fit with the boosting of back-propagation(BP)networks, the standard binary AdaBoost algorithm is extended to a multiclass version. Besides, this system adopts a hierarchical recognition framework consisting of pre-decision and post-decision. The former can roughly classify the actions into several subcategories by analyzing the locations of motion salient regions, whereas the latter exploits extra features to further enhance recognition accuracy. The experimental results on Weizmann and KTH datasets show that neural networks exhibit obvious advantages over the popular support vector machine. The BP-AdaBoost algorithm combined with hierarchical recognition can greatly reduce the computational cost and confusions among actions to achieve high recognition accuracy.

参考文献/References:

[1] Poppe R. A survey on vision-based human action recognition [J]. Image and Vision Computing, 2010, 28(6): 976-990.
[2] Efros A A, Berg A C, Mori G, et al. Recognizing action at a distance [C]//Proceedings of 2003 IEEE International Conference on Computer Vision. Nice, France, 2003: 726-733.
[3] Lertniphonphan K, Aramvith S, Chalidabhongse T H. Human action recognition using direction histograms of optical flow [C]//Proceedings of 2011 International Symposium on Communications and Information Technologies. Hangzhou, China, 2011: 574-579.
[4] Wang L M, Qiao Y, Tang X. Motionlets: mid-level 3D parts for human motion recognition [C]//Proceedings of 2013 IEEE Conference on Computer Vision and Pattern Recognition. Portland, OR, USA, 2013: 2674-2681.
[5] Laptev I, Marszalek M, Schmid C, et al. Learning realistic human actions from movies [C]//Proceedings of 2008 IEEE Conference on Computer Vision and Pattern Recognition. Anchorage, AK, USA, 2008: 1-8.
[6] Blank M, Gorelick L, Shechtman E, Irani M, Basri R. Actions as space-time shapes [C]//Proceedings of 2005 IEEE International Conference on Computer Vision. Beijing, China, 2005: 1395-1402.
[7] Grundmann M, Meier F, Essa I. 3D shape context and distance transform for action recognition [C]//Proceedings of 2008 International Conference on Pattern Recognition. Tampa, FL, USA, 2008: 1-4.
[8] Schuldt C, Laptev I, Caputo B. Recognizing human actions: a local SVM approach [C]//Proceedings of 2004 International Conference on Pattern Recognition. Cambridge, UK, 2004: 32-36.
[9] Yang M, Lü F, Xu W, et al. Human action detection by boosting efficient motion features [C]//Proceedings of 2009 IEEE International Conference on Computer Vision Workshops. Kyoto, Japan, 2009: 522-529.
[10] Li N, Cheng X, Zhang S, et al. Recognizing human actions by BP-AdaBoost algorithm under a hierarchical recognition framework [C]//Proceedings of 2013 IEEE International Conference on Acoustics, Speech and Signal Processing. Vancouver, Canada, 2013: 3407-3411.
[11] Junejo I N, Dexter E, Laptev I, et al. View-independent action recognition from temporal self-similarities [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(1): 172-185.
[12] Bobick A F, Davis J W. The recognition of human movement using temporal templates [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2001, 23(3): 257-267.
[13] Guo K, Ishwar P, Konrad J. Action recognition using sparse representation on covariance manifolds of optical flow [C]//Proceedings of 2010 IEEE International Conference on Advanced Video and Signal Based Surveillance. Boston, MA, USA, 2010: 188-195.
[14] Wong S F, Kim T K, Cipolla R. Learning motion categories using both semantic and structural information [C]//Proceedings of 2007 IEEE Conference on Computer Vision and Pattern Recognition. Minneapolis, MN, USA, 2007: 1-6.
[15] Yeffet L, Wolf L. Local trinary patterns for human action recognition [C]//Proceedings of 2009 IEEE International Conference on Computer Vision. Kyoto, Japan, 2009: 492-497.

相似文献/References:

[1]富煜清,顾明亮.基于神经网络的人脸主特征提取[J].东南大学学报(自然科学版),1995,25(5):118.[doi:10.3969/j.issn.1001-0505.1995.05.021]
 Fu Yuqing,Gu,Mingliang.Facial Feature Extraction Based on the Neural Network[J].Journal of Southeast University (Natural Science Edition),1995,25(3):118.[doi:10.3969/j.issn.1001-0505.1995.05.021]
[2]吴含前,周立凤,谢珏.二次剪枝算法在评论特征提取中的应用[J].东南大学学报(自然科学版),2016,46(3):513.[doi:10.3969/j.issn.1001-0505.2016.03.010]
 Wu Hanqian,Zhou Lifeng,Xie Jue.Application of secondary pruning algorithm in commentary feature extraction[J].Journal of Southeast University (Natural Science Edition),2016,46(3):513.[doi:10.3969/j.issn.1001-0505.2016.03.010]

备注/Memo

备注/Memo:
收稿日期: 2013-11-14.
作者简介: 李拟珺(1988—),男,博士生;吴镇扬(联系人),男,教授,博士生导师,zhenyang@seu.edu.cn.
基金项目: 国家自然科学基金资助项目(60971098)、国家自然科学基金青年基金资助项目(61302152).
引用本文: 李拟珺,程旭,郭海燕,等.基于多特征融合和分层反向传播增强算法的人体动作识别[J].东南大学学报:自然科学版,2014,44(3):493-498. [doi:10.3969/j.issn.1001-0505.2014.03.008]
更新日期/Last Update: 2014-05-20