TomoSAR Processing for Scatterers Detection using Parallel Implementation on Fast Support Detector

This article aims on describing a solution for optimizing the computational cost of a tomographic Synthetic Aperture Radar (TomoSAR) technique, using a Fast Support Generalized Likelihood Ratio Test (Fast-Sup-GLRT), for monitoring large-scale areas. The solution presented uses MapReduce framework, which has been built on parallel computing model. TomoSAR techniques are capable of estimating intensity, altitude, temporal deformation and thermal dilation of scatterers laying on the earth surface by means of multiple acquisitions of satellite-borne or airborne SAR sensors. Then, they can provide a dynamic profile of urban environment. A problem to be faced is the very high computational cost involved in the monitoring the wide areas using a large number of SAR images. The computational burden for exploring larger areas can be handled using the parallel paradigm. The experimental performance analysis on real datasets with the quantitative analysis of the computational efficiency are presented in this paper.