臺灣博碩士論文加值系統

稀疏表達在最近的追蹤演算法研究中是一項非常重要的技術，但是追蹤技術在現實生活中尚有許多需要克服的難題，例如遮蔽情況或者目標物的外型變化。在本篇論文裡，我們提出了一個以稀疏表達為基礎，能夠擷取全域性以及區域性特徵的追蹤方法。同時我們也提出一個完備的模板更新機制，能夠將目標物的變化情形持續更新到模板裡，對於全域性以及區域性的兩種模板都有獨立的更新機制。對於全域性的模板，我們使用兩個模板集合包含穩定模板以及普通模板，透過這兩種模板來獲得目標物變化的資訊，同時我們也會更新背景模板。對於區域性的資訊，我們利用更新一個由補釘影像所組成的字典來獲得最新的目標資訊。藉由這個完整的追蹤及模板更新機制，我們可以克服許多外型變化劇烈的追蹤任務。

Sparse representation is a significant technique in resent tracking research. However, there are many challenges in the real-world tracking task such as occlusion or appearance change. In this thesis, we propose a sparse-based tracking algorithm with global and local information. We also propose a robust template update scheme to catch the appearance variance. Two kinds of template are updated for global and local information independently. For global information, a stable template set and a normal template set are used to capture the appearance change. The background template set is also considered. For local information, a patches-based dictionary is updated in the tracking task. By the robust template update scheme, we can conquer serious appearance change in the tracking task.

摘要 IV
Abstract V
誌謝 VI
Table of Contents VII
List of Tables IX
List of Figures X
Chapter 1. Introduction 1
1.1 Motivation 1
1.2 Related Work 2
1.3 System Flowchart 3
Chapter 2. Object Tracking Using Particle Filter 6
2.1 State Transition Model 8
2.2 Observation Model 11
2.3 Tracking Result 12
Chapter 3. Global and Local Feature Patch Modeling for Similarity Measure in Sparsity Base 14
3.1 Sparsity-Based Discriminative Classifier (SDC) 14
3.1.1 Templates Generation 14
3.1.2 Projection Matrix S creation for Discriminative Feature Selection 16
3.1.3 Confidence Measure 18
3.2 Sparsity-Based Generative Model (SGM) 19
3.2.1 Patch-Based Dictionary D Generation Based on Target Template at 1st Frame 19
3.2.2 Patch-Based Histogram Generation 20
3.2.3 Occlusion Handling 22
3.2.4 Similarity Function 23
3.3 Collaborative Model 25
Chapter 4. Target Template Updating 26
4.1 Discriminative Model Template Updating 26
4.1.1 Positive Template Updating 26
4.1.2 Negative Template Updating 28
4.2 Generative Model Template Updating 28
Chapter 5. Experimental Results 30
5.1 Evaluate Method 30
5.1.1 Precision Plot 30
5.1.2 Success Plot 30
5.1.3 Area Under Curve (AUC) 31
5.2 Robust Evaluate Scenarios 33
5.2.1 One-Pass Evaluation (OPE) 33
5.2.2 Spatial Robustness Evaluation (SRE) 33
5.2.3 Temporal Robust evaluation (TRE) 34
5.3 Test Data 36
5.4 Experimental Results and Data Analysis 37
Chapter 6. Conclusions and Future Work 39
6.1 Conclusion 39
6.2 Future Work 40
Reference 41

[1]A. A. Argyros and M. Lourakis, “Real-time Tracking of Multiple Skin-colored Objects with a Possibly Moving Camera, Computer Vision-ECCV, Springer Berlin Heidelberg, pp. 368-379, 2004.
[2]V. Cevher, A. Sankaranarayanan, M. F. Duarte, D. Reddy, R. G. Baraniuk, and R. Chellappa, “Compressive Sensing for Background Subtraction, Computer Vision-ECCV, Springer Berlin Heidelberg, pp. 155-168, 2008.
[3]J. Gu, S. Nayar, E. Grinspun, P. Belhumeur, and R. Ramamoorthi, “Compressive Structured Light for Recovering Inhomogeneous Participating Media, Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 35, no. 3, pp. 1-1, 2013.
[4]J. Huang, X. Huang, and D. Metaxas, “Learning with Dynamic Group Sparsity, Computer Vision, IEEE 12th International Conference on, pp. 64-71, 2009.
[5]M. Isard and A. Blake, “Condensation—Conditional Density Propagation for Visual Tracking, International Journal of Computer Vision, vol. 29, no. 1, pp. 5-28, 1998.
[6]Z. Kalal, K. Mikolajczyk, J. Matas, “Tracking–Learning–Detection, Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 34, no. 7, pp. 1409–1422, 2012.
[7]J. Mairal, F. Bach, J. Ponce, G. Sapiro, and A. Zisserman, “Discriminative Learned Dictionaries for Local Image Analysis, Computer Vision and Pattern Recognition, IEEE Conference on, pp. 1-8, 2008.
[8]X. Mei and H. Ling, “Robust Visual Tracking Using L1 Minimization, Computer Vision, IEEE 12th International Conference on, pp. 1436-1443, 2009.
[9]X. Mei and H. Ling, “Robust Visual Tracking and Vehicle Classification via Sparse Representation, Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 33, no. 1, pp. 2259-2272, 2011.
[10]X. Mei, H. Ling, and D.W. Jacobs, “Sparse Representation of Cast Shadows via L1-Regularized Least Squares, Computer Vision, IEEE 12th International Conference on, pp. 583-590, 2009.
[11]D. Ross, J. Lim, R.-S. Lin, and M.-H. Yang. “Incremental Learning for Robust Visual Tracking, International Journal of Computer Vision, vol. 77, no. 1-3, pp. 125–141, 2008.
[12]Q. Wang, F. Chen, J. Yang, W. Xu, M.-H. Yang “Transferring Visual Prior for Online Object Tracking Image Processing, IEEE Transactions on, vol. 21, no. 7, pp. 3296-3305, 2012.
[13]J. Wright, A. Y. Yang, A. Ganesh, S. S. Sastry, and Y. Ma, “Robust Face Recognition via Sparse Representation, Pattern Analysis and Machine Intelligence, IEEE Transactions on, vol. 31, no. 2, pp. 210-227, Feb. 2009.
[14]Y. Wu, J. Lim, and M.-H. Yang. “Online Object Tracking: A Benchmark, Computer Vision and Pattern Recognition, IEEE Conference on, pp. 2411-2418, 2013.
[15]X. Zhang, W. Li, W. Hu, H. Ling and S. Maybank, “Block Covariance Based L1 Tracker with a Subtle Template Dictionary, Pattern Recognition, vol. 46, no. 7, pp. 1750-1761, 2012.
[16]S. Zhang , H. Yao , X. Sun and X. Lu “Sparse Coding Based Visual Tracking: Review and Experimental Comparison, Patten Recognition, vol. 46, no. 7, pp.1772 -1788 2013
[17]W. Zhong, H. Lu, and M.-H. Yang. “Robust Object Tracking via Sparsity-based Collaborative Model, Computer Vision and Pattern Recognition, IEEE Conference on, pp. 1838-1845, 2012.