即時的追蹤在電腦視覺中一直以來都是很重要的研究題目，本文裡我們提出了一個改善 Grabner和 Bischof於2006年所發表的On-line Boosting的追蹤方法。我們僅使用了簡單的Haar-like特徵來描述目標，並做篩選將表現比較好的特徵加入feature pool，而之後的強分類器訓練只需考慮這些特徵以有效的減少計算量。為了能適應前景/背景的變化，本文採取sample pool的方法進行分類器的即時更新，所賴以更新的訓練樣本則依照計算出的信任值，信任值高時更新sample pool的正樣本反之則更新負樣本；同時每隔一段時間，feature pool中的弱分類器的門檻值也會重新設定。我們並且使用background subtraction與Kalman filter來避免相似背景與雜訊的影響，並進行目標被遮蔽之後的路徑預測。實驗結果顯示，我們所提出的方法比原先的方法，能保有穩定且令人滿意的結果。

On-line tracking is main topic in computer vision. In this paper, we proposed a method based on On-line Boosting proposed by Grabner and Bischof. We only use Haar –like feature to describe target. We add good features into feature pool and then when training the strong classifier, we only use those features to decrease the computation. To adapt the changing of foreground and background we use sample pool to online update classifiers. When the confidence values are high, we update the positive samples in sample pool and vice versa. Meanwhile, we reset the threshold of weak classifiers every ten frames. We use background subtraction to avoid the effect of similar background and noise and Kalman filter to predict the object trajectory when it was occluded. From the experimental results, our method is more stable and accurate.

目錄
第一章 緒論	1
1.1 研究動機與目的	1
1.2 論文架構	4
第二章 相關文獻回顧	5
第三章 研究方法	8
3.1 Initialization	9
3.1.1 決定Region of interesting (ROI)	9
3.1.2 決定正、負樣本	10
3.1.3 Training Feature pool	11
3.1.4 弱分類器	13
3.1.5 積分影像	14
3.2 Tracking	16
3.2.1 Online Boosting	16
3.2.2 計算Confidence Map	19
3.2.3 Background Subtraction	21
3.2.4 Kalman Filter	24
3.3 Training Sample Pool and Sample Update	25
第四章 實驗結果與分析	29
4.1 與其他方法比較	31
4.2 追蹤失敗的例子	37
第五章 結論與未來研究方向	38
參考文獻	39
附錄：英文論文	41

圖目錄
圖1. 追蹤流程圖	3
圖2. 系統流程圖	9
圖3. ROI框選示意圖。	10
圖4. Training Samples. 	11
圖5. Haar-like features	13
圖6. Integral Image	14
圖7. Online boosting algorithm revised from [8]	16
圖8. Strong classifier with confidence map	20
圖9. Confidence map (modified to be shown as an image)	21
圖10. New ROI	21
圖11. 前景偵測結果	22
圖12. ROI前景	23
圖13. Confidence map with and without background subtraction	23
圖14. Positive Sample Update	26
圖15. Negative Sample Update	26
圖16. Select Negative	27
圖17. Video_1的實驗結果。	31
圖18. Video_2的實驗結果。	32
圖19. Video_3的實驗結果。	33
圖20. Video_4的實驗結果。	34
圖21. Video_5的實驗結果。	35
圖22. 失敗例子	37

[1] 	M. Dikmen, D. Hoiem and T. Huang, "A Data Driven Method for Feature Transformation," CVPR '12 Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 3314-3321, 16 06 2012. 
[2] 	J. Kwon and K. Mu, "A Unified Framework for Event Summarization and Rare Event Detection," IEEE Conference on Computer Vision and Pattern Recognition(CVPR), p. 1266 – 1273, 16-21 6 2012. 
[3] 	H. Lu, F. Yang and M. Yang, "Superpixel tracking," Computer Vision (ICCV), 2011 IEEE International Conference on, pp. 1323 - 1330, 6-13 11 2011. 
[4] 	K. Fukunaga and L. Hostetler, "The estimation of the gradient of a density function, with applications in pattern recognition," Information Theory, IEEE Transactions on (Volume:21 , Issue: 1 ), pp. 32 - 40, 1 1975. 
[5] 	G. Bradski, "Computer Vision Face Tracking For Use in a Perceptual User Interface," Intel Technology Journal Q2 ‘98, 1998. 
[6] 	B. Yang and R. Nevatia, "Online Learned Discriminative Part-Based Appearance Models for Multi-human Tracking," 12th European Conference on Computer Vision, pp. 484-498, 7-13 10 2012. 
[7] 	S. Avidan, "Ensemble Tracking," IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05)- Volume 2 , pp. 494-501, 2 2005. 
[8] 	H. Grabner and H. Bischof, "On-line Boosting and Vision," Computer Vision and Pattern Recognition, 2006 IEEE Computer Society Conference on (Volume:1 ), pp. 260 - 267, 17-22 6 2006. 
[9] 	R. Achanta, A. Shaji, K. Smith, A. Lucchi and S. Susstrunk, "SLIC Superpixels Compared to State-of-the-art Superpixel Methods," IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 2274 - 2282, 5 2012. 
[10] 	P. Viola and M. Jones, "Robust Real-Time Face Detection," International Journal of Computer Vision, pp. 137 - 154, 5 2004. 
[11] 	C. Papageorgiou, M. Oren and T. Poggio, "A General Framework for Object Detection," Computer Vision, 1998. Sixth International Conference on, pp. 555 - 562, 4-7 1 1998. 
[12] 	M. Piccardi, "Background subtraction techniques: a review," Systems, Man and Cybernetics, 2004 IEEE International Conference on, pp. 3099 - 3104 vol.4, 10-13 10 2004. 
[13] 	R. Kalman, "A New Approach to Linear Filtering and Prediction," Transactions of the ASME – Journal of Basic Engineering, pp. 35-45, 1960. 
[14] 	"OTCBVS: http://www.vcipl.okstate.edu/otcbvs/bench/". 
[15] 	"BEHAVE Interactions Test Case Scenarios: http://groups.inf.ed.ac.uk/vision/BEHAVEDATA/INTERACTIONS/". 
[16] 	S. McKennaa, S. Jabrib, Z. Duricb, A. Rosenfeldc and H. Wechslerb, "Tracking Groups of People," Computer Vision and Image Understanding, p. 42–56, 10 2000. 
[17] 	"http://www.cs.gmu.edu/~zduric/sjabri/research/". 
[18] 	"Visual Tracker Benchmark: https://sites.google.com/site/trackerbenchmark/benchmarks/v10". 
[19] 	S. Yen, J. Chien and C. Wang, "Accurate and Robust ROI Localization in CAMSHIFT Tracking Application".Multimedia Tools and Application VOLUME 71,NUMBER 3 AUGUST. 
[20] 	C. D., R. V. and M. P., "Kernel-Based Object Tracking," IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 564-577, 5 2003. 
[21] 	C. D. and M. P., "Mean shift: a robust approach toward feature space analysis," IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 603-619, 5 2002. 
[22] 	C. R.T., "Mean-Shift Blob Tracking through Scale Space," Computer Vision and Pattern Recognition, 2003. Proceedings. 2003 IEEE Computer Society Conference on (Volume:2 ), pp. 234-240, 6 2003.