臺灣博碩士論文加值系統

物件追蹤分為兩種: 單一物件追蹤與多物件追蹤。針對多物件追蹤，其主要任務是在影像串流中，區分及定位每個物件來達到標記身分和追蹤物件軌跡。常見的多物件追蹤的目標，包含街道上的行人、比賽中運動人員、行進中的車子、動物等，因物件本身外觀的高變異性，例如:光線變化、長時間性的遮擋、background clutter等，或物件外觀之間本身的相似性，使得多物件是在電腦視覺領域中相當挑戰性的研究。
常見的多物件追蹤方法中，主要解決資料間的聯結性與外觀模型建立與更新的問題。我們的研究是針對外觀學習機制藉由使用sparse coding 來改良，因為sparse coding在人臉的辨識效果相當優異。為解決外觀變異問題，提出即時性的字典學習機制，包含新增、刪除與更新三個部分，此三種機制主要幫助字典處理出現新的軌跡、離開場景畫面的軌跡與被遮蔽的軌跡。根據實驗結果，我們的方法能改善傳統ILDA (incremental linear discriminant analysis) 運算複雜的問題，時間比傳統ILDA快了5.6倍，MOTP(multiple object tracking precision)約63±5%及MOTA(multiple object tracking accuracy) 約38±10%。

Object tracking is classified into two types: single-object tracking and multiple-object tracking. In multiple object tracking the main task is to distinguish and position identified object and track their trajectories. Different objects have different forms, for example, pedestrians on a street, sports personnel in action, and moving cars and animals. However, the appearances of object are high variability caused by illumination changes, long occlusion, and background clutter. Furthermore, there are similarities in the appearances of various objects. Hence, multiple-object tracking is an important and challenging research subject in the computer vision field.
Multiple-object tracking methods are widely used to solve data association problems and establish and update model appearance. Our research is aimed at improving appearance-learning mechanisms using sparse coding because it is very effective for face recognition. To solve the problem of appearance variation, we propose an online dictionary learning mechanism that comprises the increase, delete, and update functions; these functions identify a new tracklet, leave the tracklet a given scene, and occlude a tracklet, respectively. Based on experimental results, our method performs better with complex problems than traditional improved linear discriminant analysis (ILDA) methods. The computation time is a fifth of that required by traditional ILDA methods, MOTP is 63% ± 5%, and MOTA is 38% ± 10%.

目錄
摘要 I
ABSTRACT II
誌 謝 III
目錄 IV
圖目錄 VII
一、導論 1
1.1研究動機 1
1.2研究架構 1
二、文獻探討 3
2.1 物件追蹤 (Object Tracking) 3
2.1.1 多物件追蹤-批次處理 3
2.1.2 多物件追蹤-即時處理 6
2.2 資料關聯 (Data Association) 7
2.2.1 匈牙利演算法(Hungarian Algorithm) 8
2.3 外觀模型 9
2.3.1 稀疏編碼 (Sparse Coding) 10
2.3.2 增量線性判斷分析 (Incremental Linear Discriminant Analysis, ILDA) 12
2.4 運動模型 (Motion Model) 13
2.4.1 卡爾曼濾波器Kalman Filter 14
三、系統架構 16
四、系統流程 18
4.1 行人偵測介紹 18
4.2 特徵擷取方法與更新模型 19
4.3 字典的學習 20
五、實驗 25
5-1評估方法 25
5-2. 資料庫(Database) 26
5-3 實驗環境 27
5-4 實驗結果 27
六、結論 33
七、參考 34

[1] A. Andriyenko, S. Roth, and K. Schindler, “Continuous Energy Minimization for Multi-Target Tracking,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 35, No. 1, pp.58-72, 2014.
[2] B. Leibe, K. Schindler, and L. Van Gool, “Coupled Detection and Trajectory Estimation for Multiobject Tracking,” IEEE International Conference on Computer Vision, pp. 1-8, 2007.
[3] B. Wu and R. Nevatia, “Detection and tracking of multiple, partially occluded humans by Bayesian combination of edgelet based part detectors,” In International Journal of Computer Vision, Vol. 75, No. 2, pp. 247-266, 2007.
[4] B. Yang and R. Nevatia, “An online learned CRF model for multi-target tracking,” In Computer Vision and Pattern Recognition, pp. 2034-2041, 2012.
[5] B. Yang and R. Nevatia, “Multi-target tracking by online learning of non-linear motion patterns and robust appearance models,” In Computer Vision and Pattern Recognition, pp. 1918-1925, 2012.
[6] C.H. Kuo and R. Nevatia, “How does person identity recognition help multi-person tracking?,” In Computer Vision and Pattern Recognition, pp. 1217-1224, 2011.
[7] Deep Learning web at
http://cs229.stanford.edu/materials/CS229-DeepLearning.pdf
[8] E.G. Ortiz, A. Wright, and M. Shah, “Face recognition in movie trailers via mean sequence sparse representation-based classification.” In Computer Vision and Pattern Recognition, pp. 3531-3538, 2013.
[9] F. Poiesi, R. Mazzon, and A. Cavallaro, “Multi-target tracking on confidence maps: An application to people tracking,” Computer Vision and Image Understanding, Vol. 117, No. 10, pp. 1257-1272, 2013.
[10] G. Shu, A. Dehghan, O. Oreifej, E. Hand, and M. Shah, “Part-based multiple-person tracking with partial occlusion handling,” In Computer Vision and Pattern Recognition, pp. 1815-1821, 2012.
[11] G. Welch and G. Bishop, “An introduction to the Kalman filter,” Dept. Comput. Sci., Univ. North Carolina, Chapel Hill, Tech. Rep. TR95041, 2000.
[12] J. Mairal, E. Michael, and S. Guillermo, “Sparse representation for color image restoration”, IEEE Transactions on Image Processing, Vol. 17, No. 1, 7.1, pp.53-69, 2008.
[13] J. Xing, H. Ai, and S. Lao, “Multi-object tracking through occlusions by local tracklets filtering and global tracklets association with detection responses,” In Computer Vision and Pattern Recognition, pp. 1200-1207, 2009.
[14] L. Zhang, Y. Li, and R. Nevatia, “Global data association for multi-object tracking using network flows,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 1-8, 2008.
[15] M.D. Breitenstein, F. Reichlin, B. Leibe, E. Koller-Meier, and L. J. V. Gool, “Online multiperson tracking-by-detection from a single, uncalibrated camera,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 33, No. 9, pp. 1820-1833, 2011.
[16] M. Yang, D. Zhang, and J. Yang, “Robust sparse coding for face recognition”, CVPR, pp. 625-632, 2011.
[17] N. Dalal and B. Triggs, “Histograms of oriented gradients for human detection,” IEEE Conference on Computer Vision and Pattern Recognition, Vol. 1, pp.886-893, 2005.
[18] R.E. Kalman, “A New Approach to Linear Filtering and Prediction Problems”, Journal of Fluids Engineering, Vol. 82, No. 1, pp.35-45, 1960.
[19] S.H. Bae and K. J. Yoon, “Robust Online Multi-Object Tracking based on Tracklet Confidence and Online Discriminative Appearance Learning,” IEEE Conference on Computer Vision and Pattern Recognition, pp.1218-1225, 2014.
[20] W. Brendel, M. Amer, and S. Todorovic, “Multiobject tracking as maximum weight independent set,” In Computer Vision and Pattern Recognition, pp. 1273-1280, 2011.
[21] X. Song, J. Cui, H. Zha, and H. Zhao, “Vision-based multiple interacting targets tracking via on-line supervised learning,” In European Conference on Computer Vision, pp. 642-655, 2008.
[22] Z. Cao, Q. Yin, X. Tang, and J. Sun, “Face recognition with learning-based descriptor,” IEEE Conference on Computer Vision and Pattern Recognition, pp. 2707-2714, 2010.