近十年來，壓縮感知是信號處理領域中的重大發現之一，它能夠以極少量的信號重建出原本的信號，達到即時計算的目的。2012年 “Real-Time Compressive Tracking”成功的應用壓縮感知於視覺追蹤上，由於方法簡明計算迅速，受到許多學者的關注。他們的方法利用滿足壓縮感知理論RIP條件的隨機稀疏矩陣，對多尺度的影像特徵進行降維，接著以簡易貝氏分類器進行分類達到追蹤的目的。然而，若在追蹤的過程中持續更新模型，模型很容易於發生偏移(drift)的情況下更新錯誤的資訊導致追蹤失敗。因此，本文提出兩點改良，第一，採用動態更新模型的方式，以分類器得到的分數來決定是否模型更新而確保模型的精確性；第二，基於被追蹤的目標大多落於追蹤視窗的正中間，因此特徵與視窗中心的距離以及常被挑到特徵，都分別賦予較高的權重，再以兩者的相關性 (correlation) 選擇出最佳的一組特徵。經過大量實驗後，第一點改良確實降低drift問題的機率而增加模型的精確性，然而第二點改良沒有明顯的改進，可能是因為限制特徵挑選的方式與壓縮感知的方法產生矛盾所導致，或者是因為強調視窗中心而忽略背景所蘊含的資訊。另外，我們發現隨機選取特徵和隨機選取更新的訓練樣本導致較差的穩定性，而不同的影片所適合的特徵也有差異性，也就是特徵不具有推廣性。

The past ten years, compressive sensing is an important discover on the topic of signal processing, it can use a few signals to represent the source signals for the purpose of the real-time computation. In 2012, Zhang et al. in “Real-Time Compressive Tracking” [2] successfully applied it on visual tracking and attracted attention of researchers in computer vision field due to the simple algorithm and low computation. They use a sparse random matrix that satisfies the RIP condition of compressive sensing to reduce the dimension of the multi- scales Haar-like feature of images and build positive and negative Gaussian models for each selected feature. Then, target is localized by the naïve Bayes classifier. However, models are apt to update wrong information when the drift problem happens and the system remains updating models in the tracking. Thus, we propose an improvement to ensure the accuracy of models, which adaptively updates models depending on the score from the classifier. After a large number of experiments, we demonstrate that our improvement increases the accuracy of models by reducing the drift problem. In addition, we find out that the tracking results may be quite different due to randomness in selected features and best feature combination may not be applicable for different tracking scenarios.

目錄

第一章、	緒論	1
1.1研究動機與目的	1
1.2論文架構	3
第二章、	相關文獻	4
第三章、	研究方法	9
3.2	Compressive tracking	10
3.2.1	挑選隨機稀疏矩陣	10
3.2.2	特徵萃取	12
3.2.3	貝氏分類器的建立與更新	13
3.2.4	追蹤視窗的尺度調整	15
3.3	Improved model updating	16
第四章、	實驗結果與分析	22
4.1	實驗結果比較	23
4.2	其他實驗結果(無ground truth)	29
4.3	特徵的穩定性與推廣性	31
第五章、	結論與未來研究方向	33
參考文獻	35

 
圖目錄

圖 1. 隨機稀疏矩陣示意圖	5
圖 2. 互補的隨機稀疏矩陣示意圖	6
圖 3. 系統流程圖	10
圖 4. ρ高低示意圖	11
圖 5. Center error比較圖	12
圖 6. v_i挑選到的特徵示意圖	13
圖 7. 取樣範圍說明圖示	13
圖 8. 追蹤視窗調整比例示意圖	16
圖 9. 目標物被遮蔽	16
圖 10. 影片圖示	23
圖 11. 所有版本的中心點誤差比較	29
圖 12. 追蹤失敗之例子	34

 
表目錄

表 1. 每個影片的長度與挑戰	22
表 2. Version_1與Version_2的中心點誤差比較	25
表 3. Version_1與Version_3的中心點誤差比較	26
表 4. Version_1與Version_3的中心點誤差比較，且兩者是用相同的特徵	27
表 5. Version_1與Version_4的中心點誤差比較	28
表 6. Video: BEHAVE-06280-06490的實驗結果	30
表 7. 特徵穩定性的測試結果	31
表 8. 特徵推廣性的測試結果	32

[1] H. Li, C. Shen and Q. Shi, “Real-time visual tracking using compressive sensing,” CVPR, 2011, pp. 1305-1312.
[2] K.-h. Zhang, L. Zhang and M.-h. Yang, “Real-time compressive tracking,” Computer Vision–ECCV, 2012, pp. 864-877.
[3] Q.-p. Zhu, J. Yan, H. Zhang, C.-e. Fan and D.-x. Deng, “Real-time tracking using multiple features based on compressive sensing,” Optics and Precision Engineering, 21(2), Feb 2013.
[4] S.-l. Wang, X.-g. Xiang, “Real-time tracking using multi-features weighting based on compressive sensing,” Application Research of Computers, 31(3), Mar, 2014.
[5] H. Grabner and H. Bischof, “Online boosting and vision,” CVPR, Vol. I, 2006, pp. 260-267.
[6] E. Candes and T. Tao, “Near optimal signal recovery from random projections and universal encoding strategies,” IEEE Trans. Inform. Theory, 52, 2006, pp. 5406-5425.
[7] D. Donoho, “Compressed sensing,” IEEE Trans. Inform. Theory, 52, 2006, pp. 1289-1306.
[8] H. Grabner, M. Grabner and H. Bischof, “Real-time tracking via online boosting.” BMVC, 2006, pp. 47-56.
[9] Y. C. Pati, R. Rezaiifar and P. S. Krishnaprasad, “Orthogonal matching pursuit: Recursive function approximation with applications to wavelet decomposition,” Proc. Annual Asilomar Conf. Signals, Systems, & Computers, 1993, pp. 40-44.
[10] J. Wright, A. Y. Yang, A. Ganesh, S. S. Sastry and Y. Ma, “Robust face recognition via sparse representation,” IEEE Trans. Pattern Anal. Mach. Intell., Vol.31, 2009, pp. 210-227.
[11] R. Baraniuk, M. Davenport, R. DeVore and M. Wakin, “A simple proof of the restricted isometry property for random matrices,” Constr. Approx 28, 2008, pp. 253-263.
[12] “Visual Tracker Benchmark,” http://cvlab.hanyang.ac.kr/tracker_benchmark/datasets.html
[13] M. A. Herman and T. Strohmer, “High-Resolution Radar via Compressed Sensing,” Signal Processing, IEEE Transactions on, 57(6), June 2009, pp. 2275-2284.
[14] I. Leichter, “Mean Shift Trackers with Cross-Bin Metrics,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 34(4), April 2012, pp. 695-706.
[15] S. Avidan, “Ensemble Tracking,” Pattern Analysis and Machine Intelligence, IEEE Transactions on, 29(2), Feb 2007, pp. 261-271.
[16] D.-y. Lee, J.-y. Sim and C.-s. Kim, “Fast object tracking using color histograms and patch differences,” IEEE International Conference on Image Processing (ICIP), Sept 2013, pp. 3905-3908.
[17] X. Mei and H. Ling, “Robust visual tracking using L1 minimization,” IEEE International Conference on Computer Vision (ICCV), 2009, pp. 1436-1443.
[18] D. Wang, H.-c. Lu and M.-h. Yang, “Online Object Tracking With Sparse Prototypes,” Image Processing, IEEE Transactions on, 22(1), Jan 2013, pp. 314-325.
[19] P. Dollar, Zhuowen Tu, P. Perona and S. Belongie, “Integral Channel Features,” Proceedings of the British Machine Conference, Sept 2009, pp. 91.1-91.11.
[20] D. Achlioptas, “Database-friendly random projections: Johnson-Lindenstrauss with binary coins,” Journal of Computer and System Sciences, 66(4), June 2003, pp. 671-687.
[21] M. Lusting, D. Donoho and J. M. Pauly, “Sparse MRI: The Application of Compressed Sensing for Rapid MR Imaging,” Magnetic Resonance in Medicine, 58(6), 2007, pp. 1182-1195.