Combining modern spectral estimation with Time-Frequency representation

With Joint Time-Frequency Analysis (JTFA), one can estimate the speed of moving targets from Synthetic Aperture Radar (SAR) images. These targets are usually compact. JTFA estimation of water surface speeds is more difficult since the radar returns are weak and generated by Bragg scattering patches, randomly distributed within the radar's footprint, both in time and space. To increase the signal strength, JTFA is applied to multiple range lines, which requires a generalization of the Time-Frequency representation (TFR). To do this, modern spectral estimation (MSE) techniques, which depend on estimating the data covariance matrix, are weaved into the framework of a TFR. The resulting TFR formed along azimuth-lines naturally integrates the data across multiple range-lines. We discuss and illustrate this technique on simulated data and then apply it to estimate the speed of moving water in a single-phase center SAR image. This method also enhances the signal-to-clutter ratio (SCR).