F-SAR GMTI Processor Concept for Traffic Monitoring Applications

Abstract

The algorithm described in the paper takes into account the road network for vehicle detection and parameter estimation. The idea using a road network is not new, but up to now such a road network mainly was used together with displacement based GMTI algorithms. These algorithms directly measure the azimuth displacements of the vehicles due to conventional SAR focusing. The required preprocessing is time consuming since in general a SAR image has to be generated taking into account the full bandwidth given by the pulse repetition frequency (PRF) of the radar. The across-track velocities of the vehicles are then computed by exploiting the known relationship between azimuth displacement and across-track velocity. However, problems may arise if there are adjacent roads. In that case an ambiguity free vehicle road assignment and hence, accurate parameter estimation might become a challenging task. Furthermore, often the range displacements of the moving vehicle images after common SAR focusing are neglected. Considering typical airborne systems the range displacements of the moving vehicles are in the order of dozens of meters and therefore must not be neglected for obtaining accurate position and velocity estimates. Our proposed algorithm overcomes the just mentioned problems. Conventional SAR focussing is not required since the algorithm operates directly on the range-compressed data. As a priori information the geographical coordinates as well as the elevation of the roads to be observed are necessary. Even with a single-channel SAR system for fast moving vehicles the parameters position, absolute velocity and heading can be estimated with high accuracy in near real time. The algorithm doesn’t measure vehicle displacements but uses an innovative method for parameter estimation and geocoding of the detected vehicles. With dual- or multi-channel systems potential ambiguities can be resolved and the amount of false detections can be reduced.