Traffic flow measurement system using image processing

Abstract

This paper describes a Traffic Flow Measurement System, TMS-II, based on the vehicle cluster concept. TMS-II is a system which aims at a real-time and precise measurement of spatially distributed traffic data, such as vehicle locations, vehicle speeds, traffic congestion, queue tail points, etc., through image processing. The fundamental measurement element in TMS-II is vehicle location, from which other spatially distributed traffic data are derived. In the image processing method, TMS-II adopts both line-differential method and clustering method based on the vehicle cluster concept. By introducing the vehicle cluster concept, TMS-II flexibly detects individual vehicles under light traffic conditions, as well as vehicle clusters under heavy traffic conditions. TMS-II mainly consists of a TV camera (two-dimensional CCD image sensor), a video pre-processor, and a micro-processing system. Field experiments have been conducted and the following results have been obtained from a preliminary analysis. (i) Vehicle cluster detection is rated at more than 97 percent. (ii) It takes about 0.5 secnd to calculate vehicle cluster data. (iii) It is possible to measure spatially distributed vehicle speed and queue tail point.