用於分析都市交通之自動化道路模型建置技術

Automatic Road Modeling for Urban Traffic Analysis

施養孝(Yang-Hsiao Shih)

指導教授 ： 蘇志文

中原大學/電機資訊學院/資訊工程研究所/碩士(2017年)

https://doi.org/10.6840/cycu201700176

摘要

在本篇論文中，我們提出了一個能夠適應不同都市環境下的道路監控攝影機，自動化構築道路區域、進行車流分析與車道的建構來為不同攝影機場景建構道路模型的方法。在智慧型運輸系統中，都市環境下的交通分析一直是這塊研究領域中的重要議題，而相較於較為少見的感應式感測器，利用早已大量存在於都市各道路上的道路監控攝影機影像來進行交通分析，是相對有效率且實際的方法。在本論文中，我們利用找出影片中車輛流向改變的時機，將影片分割成含有單純交通流向較小片段，並利用統計學的方法，累積統計片段的前景與流向，之後再將相似流向片段進行整併，得到符合影片的主要道路區域與流向分布，最後利用結果擬合出車道的邊界，得到可用於之後交通分析的道路模型。實驗結果顯示出不論在單純(例如高架橋)或是複雜(例如十字路口)的都市環境下，我們所提出的方法都能夠建構出符合該攝影機情況的可靠道路模型。

關鍵字

道路模型 ； 車流分析 ； 車道建構 ； 都市環境

並列摘要

In this study, we propose a method to construct road model which contains road area, traffic flow, and lane boundary automatically. The road model adapt to different scenes of individual cameras in urban environment. The analysis of urban environment is always one of the most important issue in Intelligent Transportation Systems (ITS). It is more practical and efficient to use surveillance cameras on the road rather than traffic sensors as our video sources. In this study, we split a video into segments which contain only simple traffic flows by finding when the traffic flows change. We also perform statistics of foreground and flow orientation on every segment, and merge similar segments to acquire primary foreground and flow distribution of a video. Next we fit lane boundaries by previous results. By combining primary foreground distribution, primary flow distribution, and lane boundary, we get a robust road model for further traffic analysis. The experimental results shows that our road model can describe not only simple (e.g. viaduct) but also complex (e.g. crossroads) scenes well.

並列關鍵字

Road Model ； Flow Analysis ； Lane Boundary ； Urban Environment

參考文獻

[1] L. Xin, D. Yang, Y. Chen and Z. Li, “Traffic Flow Characteristic Analysis at Intersections from Multi-layer Spectral Clustering of Motion Patterns using Raw Vehicle Trajectory,” Intelligent Transportation Systems (ITSC), pp. 513-519, 2011.

[2] L. Song, F. Jiang, Z. Shi, R. Molina and A. K. Katsaggelos, “Toward Dynamic Scene Understanding by Hierarchical Motion Pattern Mining,” IEEE Trans. Intell. Transp. Syst., vol. 15, no. 3, June 2014.

[3] B. Lucas, and T. Kanade, “An Iterative Image Registration Technique with an Application to Stereo Vision,” Proc. of 7th International Joint Conference on Artificial Intelligence (IJCAI), pp. 674-679, 1981.

[4] G. Farneback, “Two-Frame Motion Estimation Based on Polynomial Expansion,” Proc. of 13th Scandinavian conference on Image analysis, pp. 363-370, 2003.

[5] J. Lee and A. C. Bovik, “Estimation and Analysis of Urban Traffic Flow,” IEEE International Conference on Image Processing(ICIP), pp. 1157-1160, 2009.