Traffic Monitoring with Air- and Spaceborne Synthetic Aperture Radar

Abstract

Synthetic aperture radar (SAR) is an established imaging method widely used by the remote sensing community. It enables two-dimensional wide-swath imaging of the Earth's surface, independent of sunlight illumination and, depending on the used wavelength, almost independent of the weather. With conventional single-channel SAR and SAR processing only stationary, non-moving targets are imaged correctly. In contrast, moving targets appear generally blurred and displaced from their actual position in the SAR image. The reason is the additional Doppler shift caused by the target motion. Often moving targets with low radar cross section are even invisible since they are covered by the signals of the clutter background. By extending the SAR with additional receiving antennas, separated in flight direction by a certain along-track baseline, wide area traffic monitoring becomes feasible. The additional receiving channels enable clutter suppression and, hence, the detection of moving vehicles with comparatively low radar cross section. With suitable multi-channel algorithms the actual vehicle positions and velocities can be estimated additionally. Many algorithms treated in the literature so far either enable only one-dimensional velocity estimation, or an inaccurate two-dimensional velocity estimation which is severely biased by possible target accelerations or by the required target tracking. Furthermore, especially for spaceborne SAR systems, the algorithms may lead to an unsatisfying position estimation accuracy. At the beginning of the present doctoral thesis the effects on SAR imagery caused by moving targets are investigated in detail and state-of-the-art parameter estimation methods are discussed. Based on these investigations and on the estimation methods three novel algorithms well suited for radar-based traffic monitoring are developed, discussed in detail and verified using real multi-channel SAR data. The first two algorithms are for multi-channel airborne SAR systems. Both of them enable an accurate and acceleration independent two-dimensional velocity estimation. The first algorithm does not rely on a priori knowledge and can therefore by applied for monitoring traffic on unknown paths or open areas. The second one includes a road database. This has the advantage that only the areas around the a priori known roads have to be scanned for moving targets. As a consequence, the algorithm has real-time capability. The third algorithm is suited for spaceborne traffic monitoring with two SAR platforms flying in a tandem configuration with platform separations in the order of a few seconds. For this, a novel method for two-dimensional velocity estimation was developed, which additionally enables target position estimation with very high accuracy without the need of any a priori knowledge.