在當物體之間具有頻繁的互動時，多物體追蹤會是一個挑戰性的題目。本篇論文提出一個新的多物體追蹤方法來處理物體之間的互動問題。在我們的方法中，物體的外形是取用能夠鑑別目標物體與非目標物體的特徵來代表。我們使用以區域為基礎的物件追蹤來追蹤個別物體並依此保留物體外形上的空間資訊。物體的位置是統整各區域的比重和各區域估計的物體位置得到的，而各個區域的比重則是由區域追蹤結果的好壞決定。當物體的特徵因為環境變化或者物體轉身而不再具有可信度時，我們使用以區域為基礎的特徵更新來依照各個特徵對應之區域的情形做特徵更新。實驗結果顯示我們的方法在追蹤時若遇到部份遮蔽或完全遮蔽時能有比較好的追蹤結果，並且當物體之間有互動時，物體外形上的鑑別度也比較能被保存下來。

Multiple object tracking (MOT) is a challenging topic when tracking objects with frequent interaction. In this thesis, we present a novel MOT method to handle interaction difficulties. In our method, we use distinguishable features between the target and non-target to represent the object appearance. Spatial information of object appearance is preserved by tracking every object using a region-based object tracker. The target location is calculated by the weighted sum of region’s estimated target location, where the region’s weight is obtained by evaluating the region tracking performance. When the selected features are no longer reliable due to human body movements or environmental changing, we use the proposed region-based feature updating method to update a feature according to its corresponding region’s status. Experimental results show that our method is robust when tracking objects during partial and serious occlusions, and the object’s discriminability in appearance model is well maintained when interaction among other objects occurs.

[1] R.T. Collins, Y. Liu, and M. Leordeanu, “Online Selection of Discriminative Tracking Features,” IEEE Trans. PAMI, vol. 27, no. 10, pp. 1631-1643, October 2005.
[2] Y. J. Yeh and C. T. Hsu, “Online Selection of Tracking Features Using AdaBoost,” IEEE Trans. Circuits and Systems for Video Technology, vol. 19, no. 3, pp. 442-446, April 2009.
[3] D. Reid, "An Algorithm for Tracking Multiple Targets," IEEE Trans. Automatic Control, vol. 24, no. 6, pp. 843-854, December 1979.
[4] J. Vermaak, S. J. Godsill and P. Perez, "Monte Carlo Filtering for Multi-Target Tracking and Data Association," IEEE Trans. Aerospace and Electronic Systems, vol. 41, no. 1, pp. 309- 332, January 2005.
[5] S. Joo and R. Chellappa, “A Multiple-Hypothesis Approach for Multiobject Visual Tracking,” IEEE Trans. Image Processing, vol. 16, no.11, pp. 2849-2854, November 2007.
[6] T. Zhao, R. Nevatia and B. Wu, “Segmentation and Tracking of Multiple Humans in Crowded Environments,” IEEE Trans. PAMI, vol. 30, no. 7, pp. 1198-1211, July 2008.
[7] W. Qu, D. Schonfeld and M. Mohamed, “Real-Time Distributed Multi-Object Tracking Using Multiple Interactive Trackers and a Magnetic-Inertia Potential Model,” IEEE Trans. Multimedia, vol. 9, no.3, pp. 511-519, April 2007.
[8] W. Hu, X. Zhou, M. Hu and S. Maybank, “Occlusion Reasoning for Tracking Multiple People,” IEEE Trans. Circuits and Systems for Video Technology, vol. 19, no.1, pp. 114-121, January 2009.
[9] Y. Boers and H. Driessen, "The Mixed Labeling Problem in Multi Target Particle Filtering," In Proc. of FUSION 2007 Conf., Quebec, July 2007.
[10] P. Pan and D. Schonfeld, “Optimal Particle Allocation in Particle Filtering for Multiple Object Tracking,” ICIP, 2007.
[11] W. L. Lu, K. Okuma, J. J. Little, "Tracking and Recognizing Actions of Multiple Hockey Players Using the Boosted Particle Filter," Image and Vision Computing, vol. 27, no. 1-2, pp. 189–205, January 2009.