Representation of Some Class of Time Frequency Distributions Towards Their Determination Through Remote Sensing Techniques

Abstract

This project is concerned with reconstruction problems arising in the context of radar signalling. The problem arising in remote sensing is a classical inverse problem. The aim is to reconstruct the density of the target tracked from the echo, which bears two features distance and velocity, with respect to the source (radar). Also adaptive and recursive procedures are described for signal representation. The representation is build up non-linearly through functions selected from a redundant family of waveforms in a forward "greedy" kind of manner or backward. Inspired by the work of H.Narparst [5] and L.Rebollo-Neira [14], the project will include experiments concerning the robustness of such methods in the case of prior knowledge of both the target environment and optimal support of the redundant family of waveforms (dictionary) used so as to represent the density of targets bearing these common features.