摘要
在对安全监控要求越来越高的今天,生物特征识别技术以其良好的安全性,有效性和不易取代性,受到人们的重视。而步态识别是利用人的走路姿势来进行生物识别的一种方法,可以在被监控对象毫无察觉的情况下对其进行检测。以其非接触、远距离识别的特点备受人们的青睐。
本文研究了基于机器学习和多视角信息融合的步态识别技术,主要研究了人体步态特征描述,利用机器学习进行降维,单视角下步态识别和多视角下信息融合的步态识别以及搭建在线步态识别系统等相关问题。针对采用背景减除法进行背景提取时0度视角下人体部分区域被分割成背景的问题,在步态能量图像的基础上,提出了动态步态能量图像,较好的解决了这一问题。并且在90度和45度视角下,由于动态步态能量图像关注的是人体区域相对变化的部分,避免了相对不变化的部分带来的负面影响,从而也取得了较好的识别结果。对于背包情况,提出一种动态腿部能量图像,抛弃上半身的特征信息,只利用腿部动态变化信息进行识别,也取得了较好的结果。在特征描述阶段,采用机器学习的方法进行降维,利用流行学习中的局部保留投影法(LPP),采用PCA+LPP的方法利用很少的维数保留了很好的特征。针对不同视角下人体的步态序列,利用信息融合的方法,用投票法和D-S证据理论法对不同视角的识别信息进行融合,有效地解决了单个视角对特征描述不完全的问题,在中科院步态库NPLSR和CASIA库上进行了测试,取得了较理想的结果。
在上述理论的基础上,尝试建立了在线步态识别系统,采用动态能量图像作为特征,利用局部保留投影进行降维,利用多视角融合的方法进行识别,在实际测试环境下,取得了较好的结果。
Today, the security is becomes more important, researchers take more attention to biometric recognition with the favorable security, validity and hard replace. Gait recognition attract the eyeballs of more and more researchers because of non-contact and far distance, which make use of the human walking pattern to recognition and identify, and executed backdoor.
In this paper, gait recognition technique based on machine learning and fusion of multi-view is researched. And gait feature description, dimensionality reduction with manifold learning and single view and fusion of multiple views gait recognition are the mainly content. Firstly we give a new gait feature describe method—dynamic gait energy image based on the method gait energy image, due to some parts of human body are transacted background at direction 0, and which avoid the negative effect of the non-moving parts at direction 45 and 90. When people get the knapsack, we only use the information of the leg, but throw away the information of the upside. Secondly we use the method principal component analysis (PCA) + locality preserving projections (LPP), get the better recognition result with the lower dimension. Thirdly the fusion method is used with the gait image sequence of three views (direction 0, direction 45 and direction 90). It solves the problem that feature is not enough in single view. Lastly the method above is tested in three gait databases (NPLSR, CASIA and timer shoot), and the results indicate that gait recognition using multiple views information fusion can obtain higher recognition rate than any other single view.
Meanwhile, a online gait recognition system is built, the method above is used, this is a attempt that gait recognition used in practical process, and it is a tested flat for gait recognition.
引文
[1] Haritaoglu I, Harwood D and Davis L. W~4:real-time surveillance of people and their activities[J]. IEEE Trans Pattern Analysis and Machine Intelligence, 2000, 22(8):809-830
[2] McKenna Setal. Tracking groups of people[J]. Computer Vision and Image Understanding, 2000, 80(1):42-56
[3] Karmann K and Brandt A. Moving object recognition using an adaptive background memory[J]. In:V Cappellini, Time-varying Image Processing and Moving Object Recognition. 2.Elsevier, Amsterdam, The Netherlands, 1990:136-142
[4] Stauffer C and Crimson W. Adaptive background mixture models for real-time tracking[J]. In:Proc IEEE Conference on Computer Vision and Pattern Recognition, Fort Collins, Colorado, 1999, 2:246-252
[5] Lipton A, Fujiyoshi H and Patil R. Moving target classification and tracking from real-time video[J]. In:Proc IEEE Workshop on Applications of Computer Vision, Princeton, NJ, 1998:8-14
[6] Collins R. A system for video surveillvnce anti monitoring: VSAM final report[J]. Carnegie Mellon University,Technical Report: CMU-RI-TR-00-12, 2000:85-103
[7] Meyer D, Denzler J and Niemann H. Model based extraction of articulated objects in image sequences for gait analysis[J]. In:Proc IEEE International Conference on Image Processing, Santo t3arbara, California, 1997:78-81
[8] Friedman N and Russell S. Image segmentation in video sequences: a probabilistic approach[J]. In:Proc the Thirteenth Conference on Uncertainty in Artificial Intelligence, RhodeIsland, USA, 1997: 236-241
[9] Stringa E. Morphological change detection algorithms for surveillance applications[J]. In:British Machine Vision Conference, Bristol, UK, 2000:402-411
[10] R.Tanawongsuwan and A.Bohick.Gait recognition from time-normalized joint angle trajectories in the walking plane[J]. In Computer Vision and Pattern Recognition, 2001, 11:726-731
[11] Cunado D, Nixon M, Carter J. Using gait as a biometric, via phase-weighted magnitude spectra[J]. In:Proc Lnternational Conference on Audio-and Video--bused Biometric Person Authentication,Crans-Montana, Switzerland, 1997:95-102
[12] B. Bhanu and J. Han. Individual recognition by kinemutic-based gait analysis[J]. In:Proc.16~(th) International Conference on Pattern Recognition (ICPR 2002), 2002,3:333-346
[13] Murase H, Sakai R. Moving object recognition in eigenspace representation: gait analysis and lip reading[J]. Pattern Recognition Letters, 1996, 17:155-162
[14] Huang P, Harris C, Nixon M. Human gait recognition in canonical space using temporal templates[J]. Vision Image and Signal Processing, 1999, 146(2):93-100
[15]Little J,Boyd J.Recognizing people by their gait:the shape of motion[J].Journal of Computer Vision Research,1998(2):2-32
[16]Shutler J,Nixon M,Ifarris C.Statistical gait recognition via temporal moments[J].In:Proc IEEC Southwest Symposium nn Image Analysis and Interpretation,Austin,Texas,2000:201-295
[17]Hayfron-A cquah J,Nixon M,Carter J.Automatic gait recognition by symmetry analysis[J].In:Proc International Conference on Audio- and Video-based Biometric Person Authentication,Halmst.ad,Sweden,2001:272-277
[18]Johnson A,Bobick A.A multi view method for gait recognition using static body parameters[J].In Proc International Conference on Audio- and Video-basedl3iometric Person Authentication,Halmstad,Sweden,2001:301-311
[19]C.BenAbdel kader,R.Cut ler,H.Nanda,and L.Davi s.Ei gengai t:Motion-based recognition of people using image self-similarity[J],in 3rd International Conference on Audio- and Video-Based Biometric Person Authentication,2001:284-294
[20]A.Kale,A.Rajagopalan,N.Cuntoor,and V.Kruger.Human identification using gait[J].International Conference on Automatic Face and Gesture Recognition,2002:336-341
[21]R.Collins,R.Gross,and J.Shi.Silhouette-based human identification from body shape and gait[J].International Conference on Automatic Face and Gesture Recognition,2002:366-371
[22]L.Lee and W.Grimson.Gait analysis for recognition and classification[J].International Conference on Automatic Face and Gesture Recognition,2002:155-162P
[23]BT Phong.Illumination for Computer Generated Pictures[J].Communication of the ACM,1975,8(6):311-317
[24]周金和,彭福堂。一种有选择的图像灰度化方法[J]。计算机工程 2006,32(20):198-200
[25]周长发。精通Visual C++图像处理编程(第3版)[M]。北京:电子工业出版社,2006,189-190
[26]Toby H.W Lan and Raymond S.T.Lee.A new representation for human gait recognition:motion silhouettes image(MSI)[J].Springer-verlag berlin Heidelberg 2005,612-618.
[27]Ju Han,Bir Bhanu,Individual recognition using gait energy image[J],IEEE Transaction on Pattern Analysis and Machine Intelligence.2006,28(2):316-322
[28]Foster,Jeff P.,Mark S.Nixon,Adam Prugel-Bennett.Automatic gait recognition using area-based metrics[J].Pattern Recognition Letters,2003,24,2489-2497.
[29]Wagg,D.K.,M.S.Nixon.,An automated model-based extraction and analysis of gait[J].IEEE Conference on Automatic Face and Gesture Recognition,2004,Seoul,Korea,2004,11-16.
[30]Murat Ekinci,Gait Recognition Using Multiple Projections[J].proceedings of the 7~(th)international conference on automatic face and gesture recognition,2006
[31]徐蓉,姜峰,姚鸿勋,流形学习概述[J],智能系统学报,2006,1(1):44-51.
[32]罗四维,赵连伟.基于谱图理论的流形学习算法[J],计算机研究与发展.2006,43(7):1173-1179
[33]X.He and P.Niyogi,Local P reserving Projections[J],proc.conf.advances in neural information processing systems,2003
[34]M.Belkin and P.Niyogi,Laplacian Eigenmaps and Spectral Techniques for Embedding and Clustering[J],Advances in Neural Information Processing Systems 14.Vancouver,British Columbia,Canada,2002
[35]Fan R.K.Chung,Spectral graph theory[J],Regional conference series in mathematics,number 92,1997
[36]X.He,S.Yan,Y.Hu,P.niyogi and H.Zhang,Face recognition using Laplacianfaces[J].IEEE Transaction on Pattern Analysis and Machine Intelligence 2005,27(3):328-340
[37]于仕琪,覃道亮,谭铁牛,用于评测视角,衣着和带包影响的步态识别标准(ICPR文章中文简化版),中科院自动化所模式识别国家重点试验室。
[38]韩宗昭,朱洪艳,段战胜多元信息融合[M].清华大学出版社 2006
[39]Waltz E,L inas J.M ultisensor Data Fusion[M].Boston:A rtech House,1990.
[40]Tahani H and Keller J.Information Fusion Computer Vision Using the Fuzzy Integral[J],IEEE Trans.on Syst.,Man,Cybern.,1990,20(3):733-741.
[41]David L Hall.M athematical Techniques in M ultisensor Data Fusion[M].Boston:A rtech House,1992.
[42]Bogler.Shafer2demp ster reasoning with application to multisensor target identification system [J].IEEE SMC,1987,17(6):968-977.
[43]潘震中.多传感器信息融合的谢佛-登普斯特方法[J].火力与指挥控制.1994,19(3):12-16.
[44]孙锐,孙上媛,葛云峰 基丁D-S证据理论的基本概率赋值的获取[J].现代机械.2006.4:22-23.
[45]熊卫,Dempster-Shafer证据理论及其解释,华南师范大学学报 2000,5(3):15-21.