本论文提出了一个利用二维时空图象进行交通自动监测的新方法。摄像机架设在公路上方，通过两个细缝检测窗口──垂直于道路方向的车辆检测窗和平行于道路方向的速度检测窗，便形成用于交通自动检测的二维全景图（PVI）和外极面图（EPI）。本文讨论时空图象生成和校准的方法，车辆检测的信息融合方法和速度估计的时空轨迹法。实验表明，基于低成本的硬件，可实时得到车辆计数、分类和速度计量等基本交通参数。

Automatic traffic monitoring plays an important role in the truly Intelligent Vehicle/Highway System(IVHS). Vision-based approach is promising since it requires no pavement adjustments and has more potential advantages such as larger detection areas and more flexible functions. However traffic flow raises interesting but difficult problems for image processing. The various light conditions, as the result of variety of weather, places a strong need on the robust algorithms, which require a great amount of computational power to meet the real-time operations of the traffic monitoring system. Great research efforts have been put on this topic all over the world, but most of the current commercial traffic monitoring image systems are cost expensive.In this paper we present a novel approach using 2D spatio-temporal images. The TV camera is mounted above the highway. The traffic is monitored and analyzed through two slice windows for each lane the vehicle detection window is along the 2D spatio - temporal (ST) images: the panoramic view image (PVI) and the epipolar plane image (EPI). The primary problem, The separation and counting of vehicles and identifying their class(size) and speed,is solved through analyzing these two ZD ST images. The problem of camera settings, ST image calibration and rectification, data integration in vehicle detection, accurate speed estimation, background updating are discussed in the paper.The features and advantages of the proposed ST approach are: (1) Adaptive signal selection. Only the vital information which is enough for the given tasks is selected. (2) Computational efficiency. Only a few scan lines are processed in each frame, and ST images are more generic and simple than frame images in this special application. (3) Information completeness. Narrow spatial viewing windows are compensatedby dense temporal sequences, and the partially-viewed large vehicles in a single frame can be reconstructed by using ST images. (4)Accurate speed estimation. Speed is estimated from the loci of the front and rear instead of the locations at two single instants. (5) Easy background updating. Updating of the background estimation is very important in dealing with the changing weather conditions, which can be done easily according to the information from the few scan lines in the ST image method. (6) COmpact representation.PVI is a compressed and visually explicit representation for the traffic flow. So PVI can be saved on the hard disk and takes the place of the traditional video tapes. We have built up a prototype system using the methods presented in the paper.