前車(Preceding vehicle)的追蹤在影像式自動駕駛輔助系統中是一個非常重要的問題.在一般的情況中我們都能輕易的解決,但是當遇到車子逆光或是有較大的陰影出現在車底的其中一方的情況時,就比較難處理.在本論文中,我們利用車輛的對稱性,鉛直邊緣偵測的投影來將陰影所造成的腳印(footprint)去除.
整個系統一開始是車輛底部腳印的偵測,我們必需將出現在有興趣的的區域(Region of interest)內將腳印給淬取出來.在淬取的過程中,先將白色的路標利用影像處理中的膨脹(dilation)將其去除.
在將車輛底部淬取出來後,我們依據這個資訊利用邊界區塊擷取法(bounding box extraction method)可將整台車的四個邊界正確取出.在找出車子正確的位置後,我們利用測量距離的公式,將所駕駛車輛到前方車輛的距離計算出來.由於我們這公式是使用於攝影機的光軸平行於地面的情況.但是在實際上的情形是當車子在行進時,多少會有些許的上下晃動,這時刻攝影機的光軸就不再平行於地面,此時會影響到我們測量到前車距離的精確度.故我們引用攝影機上下左右移動(camera pan and tilt)的校正公式來找出偏移的消失點.
在偵測消失點的過程中,我們兩段式的赫夫轉換(Hough transform)來求取出現頻率最高消失點在鉛直方向的座標.藉由這過程後,我們就可更正確求出到前車的距離.
最後,我們將所求出整台車的四個邊界和距離整合起來,藉著這兩個資訊就提供了車輛分類和安全規則建立的基礎.然後本系統可給予駕駛人一些建議,使行車能夠更安全.

Preceding vehicle tracking is an important task in vision-based automatic guidance system. In this thesis, we propose a method to solve the problem which is including region of interest (i.e. ROI) detection, vehicle footprint extraction, and vehicle bounding box extraction. After extracting the preceding vehicle, we effectively track the vehicles, and detect the passing car and distant car. Here, we also propose the distance measure formula. Through camera calibration process and vanishing point detection process, the formula can calculate the distance to the preceding vehicle accurately. Finally, we classify the type of the preceding vehicle. Finally, based on distance and the type of the tracked vehicle, our system may provide a valuable traffic information for the driver.

[1] Z. Sun, G. Bebis, and R. Miller, “On-road vehicle detection using Gabor filters and support vector machines,” International Conference on Digital Signal Processing, July, 2002, Greece.
[2] Hideo Araki, Kenichi Yamada, and Yasuhisa Hiroshima, “Development of rear end collision avoidance system,” Intelligent Vehicles Symposium, Proceedings of the 1996 IEEE.
[3] A. Bensrhair, M. Bertozzi, A. Broggi, P. Miche, S. Mousset, and G. Toulminet. A cooperative approach to vision-based vehicle detection. In Proc. ITSC, Japan, October 2001.
[4] Z. Sun, Miller, R., Bebis, G., DiMeo, D.,”A real-time precrash vehicle detection system,” Applications of Computer Vision, 2002. (WACV 2002).
[5] J. Miura , M. Itoh, and Y. Shirai, "Towards Vision-Based Intelligent Navigator: Its Concept and Prototype", IEEE Trans. on Intelligent Transportation Systems, Vol. 3, No. 2, pp. 136-146, 2002.
[6] Ito T.,Yamada K.,Nishioka K., “Preceeding vehicle recognition algorithm using fusion of laser radar and image processing,” Intelligent Vehicle 93’ Tokyo,.pp. 420-425,1993.
[7] N. Shimomura, K. Fujimoto, T. Oki, and H. Muro, “An Algorithm for Distinguish the Types of Objects on the Road Using Laser Radar and Vision,” IEEE Tran. On ITS, Vol.3, No.3, Sept. 2002.
[8] Yao Wang, Jörn Ostermann, and Ya-Qin Zhang,” Video Processing and Communications,” Prentice Hall, 2002 (Published September 2001)
ISBN 0-13-017547-1.
[9] T. Kato, Y. Ninomiya, I. Masaki, “Preceding vehicle recognition based on learning from sample images”, IEEE Transactions on Intelligent Transportation Systems, Vol 3, No. 4, pp. 252-260, 2002.
[10] G. S. K. Fung, N. H. C. Yung, G. K. H. Pang, and A. H. S. Lai, “Effective moving cast shadows detection for monocular color image sequences,” in Proc. of 11th International Conference on Image Analysis and Processing (ICIAP), pp. 404–409, September, 2001.
[11] J. W. Hsieh, S. H. Yu, Y. S. Chen, and W. F. Hu, “A Shadow Elimination Method for Vehicle Analysis,” Proc. ICPR 2004, Cambridge, Aug 2004.
[12] M. Betke, E. Haritaglu and L. Davis, “Multiple vehicle detection and tracking in
hard real time,” IEEE Intelligent Vehicles Symposium, pp. 351–356, 1996.
[13] Toshio Ito and Kenichi Yamada, “Preceding vehicle and road lanes recognition methods for RCAS using vision system,” Proceedings of the Intelligent Vehicles ‘95 ,pp.1229-1235,1995.
[14] Yue Wang, Eam Khwang Teoh and Dinggang Shen: “Lane detection and tracking using B-Snake”. Image Vision Comput. 22(4): pp.269-280, 2004.