Moving target motion estimation and focusing in SAR images

Abstract

The phase history for a target with general motion can be calculated fairly simply when the SAR system moves along a straight line, and when the motion is reasonably small. The resulting phase history is a sum of two terms: one term depends directly on the range motion, while the second term depends on the azimuth motion multiplied with the platform motion. The two terms cannot be separated based on the phase history alone. The resulting expressions show that the SAR image will contain signatures corresponding to changing frequencies. Accordingly, time-frequency methods may be used to analyze the signatures. An experiment was designed with such signatures in mind. A target moved along a trajectory approximately in the azimuth direction, with superimposed oscillations in the range direction. The resulting signature was an azimuth smear in agreement with the predictions. The smear was very similar to a signature corresponding to a building. Using time frequency methods, the signatures could be transformed to the time-frequency domain, where a clear difference could be seen. The moving target was essentially a linear chirp with oscillations superimposed, entirely in agreement with predictions. The building signature showed a more complex structure, indicating a combination of scattering mechanisms from the building. All were different from the moving target signature. Time-frequency methods may thus be used to analyze possible moving target signatures. The methods help to discriminate between moving and stationary targets, as well as focusing and relocating them within the image. Ultimately, the processing should contribute to increased probability of recognition and possibly detection of moving targets.