臺灣博碩士論文加值系統

我們呈現一個其基於中心移動演算法於可適性目標模型之視訊物件追蹤，其可適性目標模型能運用於區分追蹤目標物與其周圍背景；目標模型為紀錄一個相對可能性比率的特徵屬性於追蹤目標物與相鄰背景之間，並且此目標模型允許其動態地自我更新來反映追蹤目標物機率密度函數的變化。
然而在追蹤過程之中，採用能量型對數可能性比率來計算新產生特徵的信心度並且延伸本身區分特徵屬性特色來進行物體分割演算法。最後，分割演算法被設計成透過可能性比率的分佈來估計出追蹤目標物體輪廓邊點的定位。.
本研究呈現出有別於先前研究的發現並且提供視訊物體追蹤的許多細部檢驗。本追蹤演算法說明其能(1) 適當的選擇追蹤目標物的特徵 (2) 強韌且精準的追蹤目標物體 (3) 估計與維持理想物體的輪廓和(4)可適性學習追蹤目標物的新外貌。

We present a video object tracking algorithm based on a mean-shift algorithm with an adaptive target model that can distinguish a target object from its surrounding background. The target model records a relative likelihood ratio of the feature attributions between the target object and the surrounding background. Moreover, the target model is dynamically updated to reflect the changes of the probability density function of the tracking object. An energy-log-likelihood ratio test is applied to compute the confidences of new features in the video sequence. A segmentation algorithm is designed to estimate break points located on the object contour using the distribution of likelihood ratios. The experimental results demonstrate that the proposed tracking algorithm can (1) effectively discriminate target object from the surrounding background, (2) robustly and accurately track target object, (3) estimate and conserve object contour, and (4) adaptively learn the new appearance of target object.

1. Introduction ……………………………………………………. 1
1.1 Motivation and Goal ……………………………………………………… 1
1.2 Challenges and Strategies …………………………………………………. 3
1.2.1 Why is it difficult to tracking object ………………………………… 3
1.2.2 How to handle the difficulties ………………………………………. 4
1.3 Notation …………………………………………………………………….. 6
1.4 System Architecture ………………………………………………………....7
1.5 Thesis Overview ………………………………………….…………….8

2. Related Works..………………………………………………...…….9
2.1 Background Modeling Method ……………………………….......………..10
2.2 Flexible Template Models Method …………………. ….………….………11
2.3 Mean-Shift Method …………. …………………………………….………13

3. Discriminate Feature Selection …….………………………………14
3.1 Discriminative Feature Selection …………………………………………..14
3.1.1 Problem Scenario …………………………………………………...14
3.2 Energy-log-likelihood ratio test ………………………………………….…16
3.3 Gray Element …………………………………………………………….…20

4. New Feature Candidate ………………………………………….…21
4.1 Gray Region…………………………………………………………...……21
4.1.1 When does the target model need to update?......................................23
4.1.2 Where does the gray region come from?.............................................25
4.1.3 Contour Alignment………………………………………………..…27
4.2 Piecewise linear Smooth Method………………………………………...…30
4.3 Break point …………………………………………..…………………..…34

5. Experimental Results ………………………………………………36
5.1 Experimental Environment………………………………………..………..37
5.2 Experimental Design …………………………………………...……..37
5.3 Experimental Evaluation ………………………………………………...…38
5.4 Experimental Event
5.4.1 Homogenous object surface variation ……………………………....40
5.4.2 Complex background …………………………..…………………...44
5.4.3 Non-homogenous object surface …………………………..………..48
5.4.4 The symmetric movable complex background (1), (2) ……………..52
5.4.5 The asymmetric movable camera …………………………………...60
5.4.6 Practical application (1), (2)……………………………………..…..64

5.5 The table of average centroid error, confidence, and mean-shift
iteration: ……………………………………………………………….......72
5.6 Empirical scrutiny difference approaches……………………………….….74
5.6.1 Limitation of color space ……………………………………………74
5.6.2 Non-uniformly motion……………………………………………....76
5.6.3 Complete occlusion……………………………...…………………..77
5.7 Limitations of the study ……………………………………...............……79

6. Conclusion and future work ……………………………….……….80
6.1 Thesis contributions………………………………………………………..80
6.2 Future work ………………………………………………………………..82

Bibliography…………………………………………………………….83

[AB] Dumitras, A. Haskell, B.G., “A background modeling method by texture replacement and mapping with application to content-based movie coding.”
[B] G.R. Bradski, “Computer vision face tracking as a component of a perceptual user interface”, In workshop on Applications of Computer Visionm, pp 214-219, Princeton, NJ, 1998.
[CET] T. F. Cootes, G. J. Edwards and C. J. Taylor, "Active appearance models," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 23, pp. 681-685, 2001.
[CM]. D. Comaniciu, and P. Meer, “Mean Shift: A Robust Approach Toward Feature Analysis”, IEEE Tran. Pattern Analysis and Machine Intelligence,vol 24, no 5, pp. 603-619, May 2002.
[CTCG] T. F. Cootes, C. J. Taylor, D. H. Cooper and J. Graham, Active Shape Models - Their Training and Application., Computer Vision and Image Understanding, Vol. 61, No. 1, January, pp. 38-59, 1995.
[CYL] R.T. Collins, Yanxi Liu, and M. Leordeanu, “Online Selection of Discriminative Tracking Features”, IEEE Tran. Pattern Analysis and Machine Intelligence, vol 27, no 10, October 2005.
[CGPP]R. Cucchiara, C. Grana, M. Piccardi, and A. Prati, “Detection Moving Objects, Ghosts, and Shadows in Video Streams”, IEEE Tran. Pattern Analysis and Machine Intelligence,Vol. 25, No. 10, pp. 1337-1342, October 2003.
[CRM] D. Comaniciu, V. Rameshm, and P. Meer, “Kernel-Based Object Tracking”, IEEE Tran. Pattern Analysis and Machine Intelligence,vol 25, no 5, 2003
[GSRL] W. Grimson, C. Stauffer, R. Romano, and L. Lee, “Using adaptive tracking to classify and monitor activities in site,” Proc. IEEE Conf. on Computer and Pattern Recognition, Santa Barbara, CA, U.S.A,. 1998,pp22-29.
[ETC] G. J. Edwards, C. J. T. Taylor and T. F. Cootes, "Interpreting face images using active appearance models," presented at Third IEEE International Conference on Automatic Face and Gesture Recognition, Nara, Japan, 1998.
[JFE] A.D. Jepson, D.J. Fleet, T.F. El-Maraghi, “Robust Online Appearance Models for Visual Tracking”, Proceedings of the International Conference on Computer Vision and Pattern Recognition, Vol 1, 2001
[HR]Pakorn KaewTrakulPng, Richard Bowden, “A real time adaptive visual surveillance system for tracking low-resolution colour target in dynamically changing scenes,” Proceedings of the Int’l on Imaging Science, Systems, and Technology, Las Vegas, Nevada, June 1999, pp. 227-232.
[NCCRT]A.Neri, S. Colonnes, G. Russo, and P. Talone, “Automatic moving object and background separation,” Using Panoramic Video,“ Signal Processing, vol 66, 1998,pp 219-232.
[NML]Mircea Nicolescu, Gérard Medioni, Mi-Suen Lee, “Segmentation, tracking and iterpretation using panoramic video,” IEEE Workshop on Omnidirectional Vision, Hilton Head, South Carolina, U.S.A., June 2000
[RFG]Alexandre R.J. François and Gèrard G. Medioni, “Adaptive color background modeling for real- time segmentation of video streams,” Proceedings of the Int’l on Imaging Science, Systems, and Technology, Las Vegas, Nevada, June 1999, pp. 227-232.
[TP] M. Turk and A. Pentland, “Face recognition using eigenfaces,” Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 1991, pp. 586-591.
[SG] C. Stauffer and W.E.L. Grimson, “Learning Patterns of Activity Using Real-Time Tracking”, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. 17, no. 8, June2001.
[VPZ]J. Vass, K. Palaniappan, and X. Zhunag, “Automatic spatio-temporal video sequence segmentation,” Proc. IEEE Intl’ Conf. On Image Processing, Chicago, IL, U.S.A., 1998
[ZB] Zoran. Zivkovic, Ben. Krose, “An EM-like algorithm for color-histogram- based object tracking”, IEEE Conference Computer Vision Pattern Recognition, Vol.1, pp. I-798-I-803. 2004.