影像式偵測器於路口交通參數之擷取研究

A study on traffic parameter extraction from image deterctor at intersection

陳彥政(Yen-Cheng Chen)

指導教授 ： 張堂賢

國立臺灣大學/工學院/土木工程學研究所/碩士(2008年)

https://doi.org/10.6342/NTU.2008.02756

摘要

路口監控和資料蒐集在交通上是非常重要的課題。以往我們必須靠人工調查方式取得交通參數，在近年來，愈來愈多人研究如何發展影像偵測系統用來蒐集交通參數，如車流量、速度等。蒐集這些交通參數可以讓交通專業人員了解路況並進一步的進行交通控制。影像偵測器在快速道路的應用已經相當的成熟，用在擷取交通參數上有非常高的準確率。但是在路口上的研究則非常的少。因為路口的交通環境比快速道路還要複雜，車輛很難被良好的追蹤，但是路口的交通參數往往是很重要的，發展一套系統能適用於路口是必須的。
在台灣，機車佔有車輛非常大的比例。由於機車具有高機動性，動線複雜，使它非常難在路口中被偵測到。它成了影像偵測器擁有高信賴、高準確率的一個挑戰。
本研究中發展出一個擁有強健的演算且具有高準確率的影像式偵測器系統，經過不同情況下的測試，可達到八成以上。

關鍵字

車輛擷取 ； 動態門檻值 ； 陰影移除 ； 遮蔽處理 ； 車輛追蹤

並列摘要

Intersection Monitoring and data collection are very important issue to traffic. In recent year, more and more people research how to develop an image detector system to collect traffic parameters such as traffic flow, speed etc. To obtain these traffic parameters can improve urban traffic by optimizing signal at intersection. The research of image detector on highway had been developed maturely that can extract traffic parameter with high accurate rate. However, the researches about data collection at intersection are few. Because intersection is more complex than highway, vehicle is difficult to be tracked well at intersection. A system that can be applied at intersection has to be developed.
In Taiwan, motorcycle occupies numerous percentages of vehicles. Because of some characteristics of it, motorcycles are difficult to detect at intersection. It becomes a challenge for image detector system to have reliable ability of vehicle detection.
In this thesis, a system which has reliable detection ability and robust algorithm on real time for data collection is necessary. After testing this system, average accurate rate of it can reach more than 80%.

並列關鍵字

vehicle extraction ； dynamic threshold ； shadow removal ； occlusion resolution vehicle tracking

參考文獻

[13] Gao, J.Y., ”Traffic Information Integrated System Based on Real-Time Image Processing,” National Taiwan University of Science and Technology Taipei Taiwan, Master dissertation, 2005. (In Chinese).

[1] Badenas J. and Pla, F., ”Segmentation Based on Region-Tracking in Image Sequences for Traffic Monitoring,” Pattern Recognition, Proceedings of the 14th International Conference, Vol. 2, pp.999, 1998.

[2] Bas, E.K., Crisman, J.D., “An easy to install Camera Calibration for Traffic Monitoring,” Intelligent Transportation System, IEEE Conference, pp. 362-366, 1997.

[3] Bartolini, F., Cappellini V. and Giani, C. “Motion Estimation and Tracking for Urban Traffic Monitoring,” Image Processing, Proceedings of International Conference, Vol. 3, pp 787-790, 1996.

[7] Cheng, S.C., “Pedestrian and Vehicle Classification for Surveillance System,” National Cheng Kung University, Tainan, Taiwan, Master dissertation, 2006.