Compressed Sensing: OMP2D Based Algorithm for Fully Polarimetric ISAR

One of the main issues that arises when using CS techniques in real-time applications is related to the high computational burden and the memory footprint. In the case of separable sampling, the 2D-OMP algorithm [1] provides a consistent solution, achieving a considerable reduction in both computational complexity and memory occupation. In this document an extension of the 2D-OMP algorithm is proposed to cope with multiple polarimetric channels. For this purpose, an Objective Function, to weight data collected from all the available channels, has been introduced in the OMP2D algorithm. By proceeding in this way, decisions are taken by exploiting an increased amount of measures (and information), and they result to be more accurate and reliable. The simulations results show how the proposed algorithm (OMP2D-FP) outperforms the simple parallelization of the conventional algorithm (OMP2D), in terms of False Alarm and Detection probability (ROC curves) as well as in terms of Mean Square Error between the recovered and the observed scenes.