A vehicle classification system based on microwave radar measurement of height profiles

• Published in: 2002 International Radar Conference pp. 409 - 413
• Main Authors: Urazghildiiev, I.R, Ragnarsson, R, Wallin, K, Rydberg, A, Ridderstrom, P, Ojefors, E
• Format: Conference Proceeding
• Language: English
• Published: London IEE 2002
• Subjects: Applied sciences; Detection, estimation, filtering, equalization, prediction; Exact sciences and technology; Information, signal and communications theory; Microwave measurement techniques; Optical, image and video signal processing; Radar theory; Radiolocalization and radionavigation; Signal and communications theory; Signal, noise; Spatial variables measurement; Telecommunications; Telecommunications and information theory; distance estimates; vehicle shape; image classification; microwave radar sensor; vehicle feature vector; microwave measurement; distance measurement; road vehicles; feature extraction; radar imaging; radar antenna; image resolution; vehicle radar image; vertical radar beam; vehicle classification system; microwave radar measurement; vehicle classification algorithm; time series; sensor-vehicle distance; high-resolution measurement; narrow radar beam; signal correlation; road surface; electromagnetic wave scattering; height profile measurement; radar resolution; scatterers; radar image pre-processing; Automatic classification; Microwave detectors; Road vehicle; Road traffic; Radar signal processing; Height measurement; Remote sensing by radar; Shape detection; Target detection
• ISBN: 0852967500; 9780852967508
• ISSN: 0537-9989
• DOI: 10.1049/cp:20020318

We consider a new system for classification of vehicles. The proposed system uses the vehicle shape as measured by a microwave radar sensor as the main feature for classification. The sensor is mounted above the road surface (perpendicular to the ground plane), so that vehicles pass vertically below the sensor. The surface of a typical vehicle can be approximated as a finite number of planes, most of which (such as roof, bonnet and body) are nearly parallel to the ground plane. When a narrow radar beam from the sensor illuminates the vehicle from above, all the relevant scatterers on the vehicle surface are at approximately the same distance to the antenna and they can be lumped together as one dominant scatterer. Using signal correlation techniques, a high-resolution measurement of the sensor-vehicle distance is repeatedly performed during vehicle passage through the vertical radar beam. This time series of distance estimates forms the vehicle radar image a, which is used to create the vehicle feature vector to be used as input to the vehicle classification algorithm.