智慧型運輸系統(ITS)為車輛的安全開啟了一個新的紀元，今日的駕駛輔助系統已可預防肇事的發生。自動駕駛輔助系統可預先警告駕駛人不當的駕駛行為。適應性速度控制可確保行車速度於安全範圍，避免碰撞；近年來有多項有效又即時的方法已應用在現實生活中。現有的技術可以大致區分為雷達、雷射等基於硬體測距之偵測，動態偵測以及機器視覺偵測。其中以雷達為基礎的技術雖然擁有準確的深度訊息和較長的檢測距離，但昂貴的硬體成本造成實現的門檻過高，至少短時間內尚無法達到普及化的程度；基於物體動向偵測的技術需要付出很多努力消除環境躁聲，倘若環境的背景複雜將造成頗高的誤判機率；基於機器學習的技術實現成本低且具有較好的穩健性，隨著越來越多投入機器視覺的研究，機器視覺在多項領域中正逐步邁向成熟。
本論文提出一個基於機器視覺的即時偵測並追蹤車輛的方法，可廣泛用於各種環境中。首先偵測環境中的車輛，我們利用基於機器學習的演算法和級聯式架構在複雜動態的場景中即時偵測車輛，並提出一個結合光流演算法和卡爾曼濾波器的追蹤機制對所偵測的車輛進行追蹤。
我們針對不同環境進行實驗，如高速公路、市郊和市區街道。我們提出的方法可以有效的偵測到不同距離的車輛，在高速公路的環境上能達到93%的偵測率，而在複雜且雜訊程度頗高的環境下也能達到85%的偵測率；而所提出的追蹤機制在各種環境中亦達到一定的穩健性。整體運算速度可達每秒18~25之畫面更新率。

Intelligent transportation system (ITS) has brought a new era for safety driving. The driving assistance system prevents accidents from happening. When inappropriate driving behavior is detected, the auto driving assistant system is able to warn the drivers. Adaptive speed control ensures a safe driving speed that prevents car crash, and there are some effective and real-time approaches available. Currently, these technologies can be categorized as radar-based and laser-based distance measuring methods. And there is also motion-based detection and computer vision based detection. Currently, the laser-based technology is the best way both in fetching depth information and has longer detection distance, but the cost is too high to implement, therefore is hard to be available to public. The motion detection method is challenging when dealing with environmental noise, where the misdetection rate will rise if the background is complex. In recent year, the machine learning based technology is both low in cost and robust to noise, and there are more researches in this area, and the machine learning method is matured in many fields in computer science.
In this thesis, we propose machine learning based real-time vehicle detection and tracking method, which can be used in several of environments. First we detect vehicles in environments, and we use machine learning algorithms and cascading architectures to detect vehicles in complex scenarios with dynamic backgrounds, and we combine the optical-flow and Kalman filter tracking mechanism to track the detected vehicles.
We made our experiments in many scenarios, such as freeway, suburban and city roads. Our proposed method is able to detect vehicles in different distances. In freeway scenario, the detection rate is above 93%, whereas in the complex environments we also reach 85% detection rate. The proposed tracking method is robust in several environments. The overall performance is 18~25 fps.

[1] M. Everingham, A. Zisserman, C. Williams and L. V. Gool. The Pascal Visual Object Classes Challenge 2006 (VOC2006) Results.
[2] P. Viola and M. J. Jones, “Robust real-time object detection,” Second International Workshop on Statistical and Computational Theories of Vision, July 2001.
[3] Herbert Bay, Andreas Ess, Tinne Tuytelaars, Luc Van Gool, “SURF: Speeded Up Robust Features”, Computer Vision and Image Understanding (CVIU), Vol, 110, No. 3, pp. 346-359, 2008.
[4] Arrospide, J.; Salgado, L.; Nieto, M. “Vehicle detection and tracking using homography-based plane rectification and particle filtering, ”Intelligent Vehicles Symposium (IV), IEEE, pp. 150-155 2010.
[5] Pantilie, C.D.; Nedevschi, S. “Real-time Obstacle Detection in Complex Scenarios Using Dense Stereo Vision and Optical Flow,” Intelligent Transportation Systems(ITSC),13th International IEEE Conference, pp. 439-444, 2010.
[6] Young-Chul Lim, Chung-Hee Lee, Soon Kwon, Jonghwan Kim, “Event-Driven Track Management Method for Robust Multi-Vehicle Tracking, ” Intelligent Vehicles Symposium(IV), IEEE, pp, 189-194, 2011.
[7] Wei Zheng, Luhong Liang ,” Fast Car Detection Using Image Strip Features, ” Conputer Vision and Pattern Recognition, CVPR, IEEE Conference, pp.2703-2710, 2009.
[8] Zhang Yang, Liu Weiming, Mo Chen, Li Zilong , “Pedestrian Detection in Complex Scene Using Full Binary Tree Classifiers based on Locally Assembled Binary Haar-like features, ” Intelligent Control and Automation (WCICA), IEEE Conference, pp. 1180-1184, 2011.
[9] P. Viola and M. J. Jones, “Robust real-time object detection,” Second International Workshop on Statistical and Computational Theories of Vision, July 2001.
[10] Bo Wu and Nevatia R, “Cluster Boosted Tree Classifier for Multi-View, Multi-Pose Object Detection, ” IEEE 11th International Conference, pp. 1-8 2007.
[11]C. P. Papageorgiou, M. Oren and T. Poggio, “A general framework for object detection,” Sixth International Conference on Computer Vision, pp. 555-562, Jan. 1998.
[12] Y. Freund and R. E. Schapire, "A decision-theoretic generalization of on-line learning and an application to boosting," Journal of Computer and System Sciences, vol. 55, no. 1, pp.119-139, Aug. 1997.
[13] B.D. Lucas and T.anade, “An iterative image registration techniaue with an application to stereo vision,”Proceedings of the 1981 DARPA Imaging Understanding Workshop, pp. 121-130, 1981.
[14] Bouguet J. Y. “ Pyramidal implementation of the lucas kanade feature tracker: description of the algorithm. ,” Intel Corporation Microprocessor Research Labs. 2000.
[15] C. Harris and M. Stephens, “A combined corner and edge detector,” Proceedings of the 4th Alvey Vision Conference, pp. 147-151, 1988.
[16] J. Shi and C. Tomasi, “Good features to track,” 9th IEEE Conference on Computer Vision and Pattern Recognition, June 1994.
[17] G. Welch and G. Bishop, “An introduction to the Kalman filter,” Technical Report TR95-041, Department of Computer Science, University of North Carolina, Chapel Hill, United States, April 2004.
[18] Herbert Bay, Andreas Ess, Tinne Tuytelaars, Luc Van Gool, “Speeded-Up Robust Features (SURF), ” Computer Vision and Image Understanding 110, pp. 346-359, 2008.
[19] Jazayeri, A.; Hongyuan Cai; Jiang Yu Zheng; Tuceryan, M. “Vehicle Detection and Tracking in Car Video Based on Motion Model,” IEEE Transactions on Intelligent Transportation Systems, pp. 583-595, 2011.
[20] Jianqiang Wang, Zhifeng Liu, Shichun Yi, Keqiang Li, "Target vehicle selection based on multi features fusion method,” Intelligent Vehicles Symposium (IV), IEEE , pp.13-19, 2010.
[21] You Li, Ruichek, Y. “ Moving objects detection and recognition using sparse spatial information in urban environments,” Intelligent Vehicles Symposium (IV), IEEE , pp. 1060 – 1065, 2012.
[22] Xin Liu, Zhenping Sun, Hangen He. “ On-road Vehicle Detection Fusing Radar and Vision,” IEEE International Conference on Vehicular Electronics and Safety(ICVES), pp. 150-254, 2011.
[23] Cheng-Hao Kuo; Nevatia, R. “Robust multi-view car detection using unsupervised sub-cateforization,” Applications of computer vision (WACV), pp. 1-8 2009.