視覺式交通監測系統

Vision-Based Traffic Measurement System

王俊明(Wang Jung-Ming)

指導教授 ： 陳世旺

國立臺灣師範大學/教育學院/資訊教育學系/碩士(2002年)

摘要

道路資訊的蒐集，在許多的交通運輸應用中扮演著重要的角色；目前政府交通單位在各種道路上所裝設之攝影機，其所拍攝到的影像在傳回交控中心後，均由人工監看；由於一個人通常需兼顧多部攝影機，難免造成疏失，而且也無法即時計算交通數據。本論文針對交通監測影像，提出能自動從影像中擷取交通資訊的技術，主要目的在於利用現有的交通攝影設備，蒐集道路上之車流量及車行速度等資訊，以應用於交通監控系統。在擷取交通資訊的過程中，主要可以分成三步驟：車輛偵測、車輛追蹤及交通資訊擷取。首先從輸入的影像序列中，以一種結合建構及更新的方法產生背景影像，之後將建構出來的背景影像與目前影像作比較後，快速偵測出前景區塊；另外還利用光線估測的方法評估影像中是否含有陰影，若其含有陰影，則執行陰影移除的工作以擷取出車輛影像。在偵測出車輛後，便利用車輛本身及其互相之間的屬性作比對，以追蹤連續影像中的車行路徑及車輛數。接著利用攝影機的校對的結果，將車行資訊轉換成真正的空間位移，以計算出車行速度。最後由車輛數及車行速度推算出各項交通參數值，以擷取所需之交通資訊。我們以家用攝影機及交通單位所提供的監測影像驗証所提出的方法，證明其不但能適用於各種交通狀況，還能相當正確的擷取出交通資訊。

關鍵字

交通監測系統 ； 漸進式背景影像建構 ； 車道標線偵測 ； 攝影機校對 ； 明亮度評估 ； 陰影偵測與移除 ； 交通參數 ； HASH(0xa68938c)

並列摘要

In this paper, a vision-based traffic measurement system is addressed. The objective of this system is twofold, automatically count the vehicles passing through a roadway and measure their speeds. The collected data will be used to compute a number of traffic parameters, including flow rate, demand, time headway, throughput, mean speed, and density. This system consists of one off-line module (preprocessing) and three on-line modules (vehicle detection, vehicle tracking, and calculation of traffic parameters). In the preprocessing component, four off-line tasks are performed, background image generation, lane detection, vanishing point detection, and camera calibration. The background image is used by the vehicle detection module to quickly extract foreground objects from video images through image subtraction. Foreground objects are tracked across the video sequence by the tracking module. Once the traces of vehicles are determined, the trace count gives the number of vehicles. The 2D traces are converted into 3D ones using the transformation functions determined during camera calibration. The speeds of vehicles are then calculated. Based on the number of vehicles and their speeds, the desired traffic parameters are computed. We present the results of some experiments using real image sequences made using a home video camera and a traffic surveillance system. Finally our conclusions show that our proposed system is both adaptable and accurate.

並列關鍵字

Traffic measurement system ； Progressive background generation ； Lane detection ； Camera calibration ； Illumination assessment ； Shadow detection and elimination ； Steerable filters ； Fuzzy-set theoretic multi-user decision

參考文獻

[Chu01] J. H. Chun, Automatic Traffic Monitoring System, MS Thesis, Dept. of Information and Computer Education, National Taiwan Normal University, 2001.

[Alt98] Y. Altunbasak, P. E. Eren and A. M. Tekalp, Region-Based Parametric Motion Segmentation Using Color Information, Graphical Models and Images Processing, vol. 60, no.1, pp.13-23, 1998.

[Bad98] J. Badenas and F. Pla, Segmentation Based on Region-Tracking in Image Sequences for Traffic Monitoring, Proc. of 14th Intl Conf. on Pattern Recognition, vol. 2, pp. 999 1001, 1998.

[Bey97] D. Beymer, P. McLauchlan, B. Coifman, and J. Malik, A Real-Time Computer Vision System for Measuring Traffic Parameters, Proc. of IEEE Conf. on Computer Vision and Pattern Recognition, pp. 495 501, 1997.

[Bur01] J. B. Burns, Detecting Independently Moving Objects and Their Interactions in Geo-Referenced Airborne Video, Proc. of IEEE Workshop on Detection and Recognition of Events in Video, pp. 12 19, 2001.