智慧型公路系統中以視覺基礎之可行駛空間偵測與模型基礎之交通車流特徵化研究

Vision-Based Drivable Space Detection and Model-Based Traffic Flow Characterization in Intelligent Highway System

許志明(Chih-Ming Hsu)

指導教授 ： 連豊力

國立臺灣大學/電機資訊學院/電機工程學研究所/博士(2013年)

https://doi.org/10.6342/NTU.2013.10885

摘要

本論文提出在智慧型公路系統中以視覺基礎之可行駛空間偵測與模型基礎之交通車流特徵化研究。對於智慧型車輛與移動式機器人而言，交通場景的認知是相當重要的一個研究議題，特別在複雜且動態的環境中，去決定出可行駛的空間與可能的障礙物位置是路面場景認知中最基本的一項工作。本論文中，我們利用視覺的方法以路面色彩特徵結合路面幾何資訊，在動態的環境中感知道路與移動的障礙物，此方法主要是藉由識別相鄰區域的亮度差異性，分割出未被佔用的路面區域，此鄰域相似性的識別是一種利用統計特徵分析與結合廣度優先搜尋機制的搜尋演算法。路面模型和路面候選區域的相似性量測，則是以巴塔恰里雅距離來計量，在識別出未被佔用的路面區域之後，利用我們提出的障礙物掃描機制與攝影機線上校準法則可以估測出前方障礙物的相對距離。經由廣泛的實驗，呈現出本方法在實際的交通場景中，可偵測出可行駛區域與動態估測出前方障礙物相對距離。
在特徵化交通車流行為的研究上，為了能預測未來車流容量之變化，我們首先建立一個正確的交通車流模型，然後，以交通車流的動態特徵化與交通訊息量測定法則，分別鑑別出車流的動態變化之行為特性與量化交通動態變化之訊息。
為了精確地特徵化交通車流，我們提出了一種階層式高斯混合模型的架構，以建構出正確的經驗性動態模型。首先為了能特徵化相關時間性與地域性的參數和降低交通資料量，交通車流資料可以由多個高斯函數的線性組合來呈現，同時，為了檢驗動態變化的行為特性，我們採用相位轉變法鑑別出多種的交通車流型態和動態切換的行為特性；此外，我們收集不同的車輛偵測器的交通資料，藉以訊息熵的計量結果，可以決定出不同位置之車輛偵測器的重要性。經由收集六個月的台灣高速公路區段的交通資料與實驗分析，鑑別出五種交通車流型態，每種交通車流型態顯示出各別的特殊交通動態行為的獨特解釋意義。

關鍵字

路面偵測演算法 ； 相似性量測 ； 攝影機校準 ； 高斯混合模型 ； 訊息熵

並列摘要

The vision-based drivable space detection and model-based traffic state estimation systems are presented in this dissertation. Traffic scene understanding and perception is an important issue for intelligent vehicles and autonomous mobile robots. Especially in complex and dynamic environments, the determination of drivable space and moving obstacles are fundamental tasks for road scene understanding. In this dissertation, a vision-based approach of combining road geometry and color features to percept road and moving obstacles in a dynamic environment is proposed. In the approach, a free road surface is detected and segmented by identifying the intensity difference among neighboring regions. The identification of region similarity is one type of searching algorithm by using statistical feature analysis (SFA) combined with a breadth-first search (BFS). Then, the similarity between the road model (its color distribution) and the road region candidates is expressed by a metric derived from the Bhattacharyya distance. After the identification of the free road surface, the relative distance of preceding obstacles can be easily estimated using the obstacle scanning mechanism (OSM) and camera calibration scheme. Extensive experiments have been performed to demonstrate that the proposed approach can properly detect the drivable region and dynamically estimate the relative distance of preceding obstacles in real traffic scenes.
In characterizing traffic flow behavior, the static analysis of traffic flow is first presented for properly constructing traffic flow models for predicting future traffic volumes. Then, the dynamic characterization of the traffic flow and traffic information measurement is proposed to identify dynamically changing behaviors and to quantize the information of traffic dynamics.
To accurately characterize traffic flow, a hierarchical Gaussian mixture modeling (HGMM) framework is proposed for constructing a proper empirical dynamics model. The traffic flow data are first represented by a linear combination of multiple Gaussian functions for characterizing related timing and geographical parameters and for reducing the quantity of collected traffic data. To further examine dynamically changing behaviors, the phase-transition approach is used for identifying various traffic flow patterns and their dynamic switching behaviors. Furthermore, the information entropy on the traffic data collected at various vehicle detectors can be calculated for characterizing the location significance of these detectors. Detailed experimental analyses show that five types of traffic flow patterns can be identified on a six-month traffic data set from a highway section in Taiwan. Each traffic flow pattern indicates a distinct interpretation of a special dynamic traffic behavior.

並列關鍵字

Road detection ； drivable detection ； camera calibration ； Gaussian mixture modeling

參考文獻

[1] J. Shawe-Taylor, T. de Bie, and N. Cristianini, “Data mining, data fusion and information management,” in Proc. IEEE Conf. Intell. Transp. Syst.., vol. 153, no. 3, pp. 221–229, Sep. 2006.

[2] J. Zhang, F. Y. Wang, K. Wang, W. H. Lin, X. Xu, and C. Chen, “Data-Driven Intelligent Transportation Systems: A Survey,” IEEE Trans. Intell. Transp. Syst., vol. 12, no. 4, pp. 1624-1639, 2011.

[3] N. K. Kanhere and S. T. Birchfield, “A taxonomy and analysis of camera calibration methods for traffic monitoring applications,” IEEE Trans. Intell. Transp. Syst, vol. 11, no. 2, pp. 441–452, 2010.

[4] C. Zhaoxue and S. Pengfei, “Efficient method for camera calibration in traffic scenes,” Electronics Letters, vol. 40, no.6, pp.368–369, March, 2004.

[5] J. M. Alvarez and A. M. Lopez, “Road Detection Based on Illuminant Invariance,” IEEE Trans. Intell. Transp. Syst, vol. 12, no. 1, pp.184-193, 2010.