Radar Imaging
Introduction
I was interested during my master II internship in radar image processing. The theme of my work was the study of multi-pass differential interferometry and particularly the DInSAR by the Permanent scatters (PS) and its improvement for its use in non-urban areas. The objective was the selection of the distributed scatterers (DS), the construction of groups of this type of scatterers and the optimization of the selection of the representatives of each class in order to establish the differential model for estimating the speed of deformation as well as the phases of errors. This work allowed me to learn about a new area of research and application which is the processing of SAR radar images and particularly radar interferometry (processing of the phase of the radar signal).
​
On a personal level, this work has allowed me to approach a new line of research whose complexity, throughout the world, is on several levels: the complexity of radar systems, the physics of radar wave-medium interaction, the concepts used, the algorithmic complexity, etc. On the application level, the results obtained complement the work done and still being done at the Image Processing and Radiation Laboratory (LTIR).
Interferometry
The interferometric technique was first used for the generation of digital terrain models (DTM), then it was extrapolated for the mapping of ground movements from several images acquired in different dates and this variant is called differential interferometry DInSAR. Their principles exploit the phase difference of two signals acquired by two sensors separated spatially by a distance called baseline by adding in the DInSAR the temporal separation of acquisitions which is translated by the temporal baseline. The constraints related to these two quantities (spatial and temporal baseline) as well as the delay introduced by the atmospheric layer, are the main limitations that have led to the emergence of very advanced techniques to reduce their effects on the accuracy of results. The technique based on permanent scatterers called 'PSInSAR' (Permanent scatterers Interferometry SAR) is one of these new techniques that is very developed for the application of DInSAR in urban environments. However, in non-urban environments, this technique is deficient. In this context, my master's degree project work was mainly focused on improving the technique of PS to make it usable in non-urban environments by introducing the concept of distributed backscatter DS ( distributed Scatterers). DS are a set of adjacent pixels with similar statistical characteristics.
​
In this work, I studied and implemented a DS selection process combined with PS selection for the resolution of the land surface deformation velocity estimation model as well as the topographic and atmospheric errors.
