臺灣博碩士論文加值系統

人體姿態的分析於行為辨識、人機介面互動及監控系統的發展及應用占有重要地位。無標記式人體姿態分析可提供非侵入式且高自由度的姿勢擷取，但因為動作自由度高且人物衣著變化大，因此其挑戰極高。本論文提出一個新式的人體姿態擷取方法，該方法針對單張影像，定位出二維影像中人體關節點位置，並還原在三維空間中的人體姿勢。本論文方法建立有向貝式網路模型，並提出以Annealing Gibbs Sampling 推論法估算該模型節點之機率，以求得人體關節於空間中的位置。實驗使用HumanEva影像資料庫來驗證提出方法的有效性，實驗人員無任何衣著限制及標記，實驗動作為高自由性的繞圈步行運動。實驗結果證明本論文提出之方法可有效的估測影像中人體的姿態。

Human pose estimation method is important for the development of behavior recognition, human-robot interaction and visual surveillance. Markerless human pose estimation method can provides non-intrusive and high-free motion capture. It has great challenges due to large range variations of motion and clothe. We propose a novel human motion capture method. The proposed method can locate human body joint position and reconstruct the human pose in 3D space from monocular images. We propose a directed probabilistic graphical model to estimate human joint positions by a devised annealing Gibbs sampling inference method. Experiments are conducted on HumanEva dataset to show the effectiveness of the proposed method. Subjects in the HumanEva have no clothe lamination and markers. The experimental data are image sequences of walking motion around a region with large ranges variation of pose. Experimental results show that the proposed method can estimate human pose from monocular images efficiently.

Abstract(in Chinese).......................................................................................................... i
Abstract............................................................................................................................ ii
Acknowledgement(in Chinese) ....................................................................................... iii
Contents.......................................................................................................................... iv
List of Figures................................................................................................................... v
List of Tables.................................................................................................................. vii
Chapter 1 Introduction...................................................................................................... 1
1.1 Background............................................................................................................. 1
1.2 Motivation .............................................................................................................. 4
1.3 Brief introduction to probabilistic graphical models.............................................. 5
1.4 System architecture ................................................................................................ 9
1.5 The organization of this thesis.............................................................................. 10
Chapter 2 Review of Previous Work.............................................................................. 11
Chapter 3 Feature Extraction.......................................................................................... 16
3.1 Human silhouette.................................................................................................. 16
3.2 Normalized center of human body ....................................................................... 17
3.3 Spatial distribution of skin color .......................................................................... 19
3.4 Corners of lower body .......................................................................................... 21
Chapter 4 Pose Estimation Framework .......................................................................... 23
4.1 Articulated human model ..................................................................................... 24
4.2 Approximate inference ......................................................................................... 28
4.3 Component-Based Metropolis-Hastings approach............................................... 30
4.4 Component-based Gibbs sampling approach ....................................................... 32
4.5 Component-Based annealing Gibbs sampling approach...................................... 36
Chapter 5 Experimental result ........................................................................................ 38
5.1 Database ............................................................................................................... 38
5.2 2D pose estimation result ..................................................................................... 42
5.3 3D pose estimation result ..................................................................................... 51
Chapter 6 Conclusion ..................................................................................................... 61
References ...................................................................................................................... 63

[1]R. Poppe, "Vision-based human motion analysis: An overview," Computer Vision and Understanding, Vol. 108, No. 1-2, pp. 4-18, 2007.
[2]A. O. Balan, L. Sigal, and M. J. Black, "A quantitative evaluation of video-based 3D person tracking," Proceedings, International Visual Surveillance and Performance Evaluation of Tracking and Surveillance Conference, Beijing, China, pp. 349-356, 2005.
[3]L. Sigal, S. Bhatia, S. Roth, M. J. Black, and M. Isard, "Tracking loose-limbed people," Proceedings, International Computer Vision and Pattern Recognition Conference, Washington, USA, pp. 421-428, 2004.
[4]L. Sigal and M. J. Black, "Predicting 3D people from 2D pictures," Proceedings, International Articulated Motion and Deformable Objects Conference, Port d'Andratx, Spain, pp. 185-195, 2006.
[5]F. V. Jensen, Bayesian networks and decision graphs, USA, Springer, 2001.
[6]C. Huang and A. Darwiche, "Inference in belief networks: a procedural guide," International Journal of Approximate Reasoning Vol. 15, No. 3, pp. 225-263, 1996.
[7]R. J. McEliece, D. J. C. MacKay, and J.-F. Cheng, "Turbo decoding as an iInstance of Pearl's “belief propagation” algorithm," IEEE Journal on Selected Areas in Communications, Vol. 16, No.2, pp. 140-152, 1998.
[8]F. R. Kschischang, B. J. Frey, and H.-A. Loeliger, "Factor graphs and the aum-product algorithm," IEEE Transactions on Information Theory, Vol. 42, No.2, pp. 498-519, 2001.
[9]L. Sigal and M. Black, "HumanEva: synchronized video and motion capture dataset for evaluation of articulated human motion," CS-06-08, Brown Univ, 2006.
[10]M. Bray, "Shadow puppetry," Proceedings, International Computer Vision Conference, Kerkyyra, Greece, pp. 1237-1244, 1999.
[11]A. M. Elgammal and C.-S. Lee, "Inferring 3D body pose from silhouettes using activity manifold learning," Proceedings, International Computer Vision and Pattern Recognition Conference, Washington, USA, pp. 681-688, 2004.
[12]G. Loy, M. Eriksson, J. Sullivan, and S. Carlsson, "Monocular 3D reconstruction of human motion in long action sequences," Proceedings, International European Computer Vision Conference, Prague, Czech Republic, pp. 139-150, 2004.
[13]R. E. Rosales and S. Sclaroff, "Inferring body pose without tracking body parts," Proceedings, International Computer Vision and Pattern Recognition Conference, Hilton Head Island, SC, pp. 721-727, 2000.
[14]E.-J. Ong and S. Gong, "A dynamic 3D human model using hybrid 2D-3D representations in hierarchical pca space," Proceedings, International British Machine Vision Conference, Nottingham, United Kingdom, pp. 33-42, 1999.
[15]C. J. Taylor, "Reconstruction of articulated objects from point correspondences in a single uncalibrated image," Computer Vision and Image Understanding, Vol. 8, No. 3, pp. 349-363, 2000.
[16]A. Agarwal and B. Triggs, "A local basis representation for estimating human pose from clustered images," Proceedings, International Computer Vision Conference, Hyderabad, India, pp. 50-59, 2006.
[17]A. Agarwall and B. Triggs, "Recovering 3D human pose from monocular images," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 28, Issue 1, pp. 44-58, 2006.
[18]R. Bowden, T. A. Mitchell, and M. Sarhadi, "Non-linear statistical models for the 3D reconstruction of human pose and motion from monocular image sequences," Image and Vision Computing, Vol. 18, Issue 9, pp. 729-737, 2000.
[19]K. Rohr, "Towards model-based recognition of human movements in image sequences," Computer Vision, Graphics and Image Processing: Image Understanding, Vol. 59, Issue 1, pp. 94-115, 1994.
[20]T. J. Roberts, S. J. McKenna, and I. W. Ricketts, "Human tracking using 3d surface colour distributions," Image and Vision Computing, Vol. 24, Issue 12, pp. 1332-1342, 2006.
[21]S.-F. Lan, M.-F. Ho, and C.-L. Huang, "Human motion parameter capturing using particle filter and nonparametric belief propagation," Proceedings, International Southwest Symposium on Image Analysis and Interpretation Conference, Santa FE, New Mexico, pp. 37-40, 2008.
[22]G. Hua, M.-H. Yang, and Y. Wu, "Learning to estimate human pose with data driven belief propagation," Proceedings, International Computer Society on Computer Vision and Pattern Recognition Conference, San Diego, CA, pp. 747-754, 2005.
[23]D. Ramanan, D. A. Forsyth, and A. Zisserman, "Tracking people by their appearence," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 29, Issue 1, pp. 65-80, 2007.
[24]M. W. Lee and I. Cohen, "A model-based approach for estimating human 3D poses in static images," IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 28, Issue 6, pp. 905-916, 2006.
[25]N. Thome, D. Merad, and S. Miguet, "Human body part labeling and tracking using graph matching theory," Proceedings, International Video and Signal Based Surveillance Conference, Sydney NSW, Australia, pp. 38-46, 2006.
[26]T. Leonid and T. Darrell, "Bayesian articulated tracking using single frame pose sampling," Proceedings, International Statistical and Computational Theories of Vision Conference, Nice, France, pp. 1-14, 2003.
[27]N. Jojic, M. Turk, and T. S. Huang, "Tracking self-occluding articulated objects in dense display maps," Proceedings, International Computer Vision Conference, Kerkyra, Greece, pp. 123-130, 1999.
[28]N. Jojic and B. J. Frey, "Learning flexible sprites in video layers," Proceedings, International Computer Vision and Pattern Recognition Conference, Kauai, Hawii, pp. 199-206, 2001.
[29]D. M. Gavrila and L. S. Davis, "Tracking of humans in action: A 3D model-based approach," Proceedings, International Computer Vision and Pattern Recognition Conference, San Francisco, CA, pp. 73-80, 1996.
[30]C. Bregler, J. Malik, and K. Pullen, "Twist based acquisition and tracking of animal and human kinematics," International Journal of Computer Vision, Vol. 56, Issue 3, pp. 179-194, 2004.
[31]S. Wachter and H.-H. Nagel, "Tracking persons in monocular image sequences," Computer Vision and Image Understanding, Vol. 74, Issue 3, pp. 174-192, 1999.
[32]P. F. Felzenszwalb and D. P. Huttenlocher, "Pictorial structures for object recognition," International Journal of Computer Vision, Vol. 61, No. 1, pp. 55-79, 2005.
[33]L. Sigal, M. Isard, B. Sigelman, and M. J. Black, "Attractive people: Assembling loose-limbed models using non-parametric belief propagation," Proceedings, International Neural Information Processing Systems Conference, Vancouver, Canada, pp. 1539-1546, 2003.
[34]D. Ramanan and C. Sminchisescu, "Training deformable models for localization," Proceedings, International Computer Vision and Pattern Recognition Conference, New York, NY, pp. 206-213, 2006.
[35]G. Mori, X. Ren, A. A. Efros, and J. Malik, "Recovering human body configurations: Combining segmentation and recognition," Proceedings, International Computer Vision and Pattern Recognition Conference, Washington, USA, pp. 326-333, 2004.
[36]X. Ren, A. C. Berg, and J. Malik, "Recovering human body configurations using pairwise constraints between parts," Proceedings, International Computer Vision Conference, Beijing, China, pp. 824-831, 2005.
[37]R. Navaratman, A. Thayananthan, P. H. Torr, and R. Cipolla, "Hierarchical part-based human body pose estimation," Proceedings, International British Machine Vision Conference, Oxford, United Kingdom, pp. 479-488, 2005.
[38]M. W. Lee, I. Cohen, and S. K. Jung, "Particle filter with analytical inference for human body tracking," Proceedings, International Motion and Video Computing Conference, Orlando, FL, pp. 159-168, 2002.
[39]C. Sminchisescu, A. Kanaujia, and D. Metaxas, "Learning joint top-down and bottom-up processes for 3D visual inference," Proceedings, International Computer Vision and Pattern Recognition Conference, New York, NY, pp. 1743-1752, 2006.
[40]S. Suzuki and K. be, "Topological Structural Analysis of Digital Binary Images by Border Following, " Computer Vision, Graphics, and Image Processing Vol. 30, No. 1, pp. 32-46, 1985.
[41]T. Lindeberg, "Feature detection with automatic scale selection," International Journal of Computer Vision, Vol. 30, No. 2, pp. 79-116, 1998.