TABLE OF CONTENTS We have seen that SAR images could be produced in radar or ground geometry: the approach that has made it possible for us to make this distinction was based on hypotheses of flat land as well as on perfect sensor knowledge. In this chapter, we will approach issues related to sensor localization and image registration.
The main geometric difference existing between radar imaging and traditional (optical) imaging is due to the fact that it only involves distances and not angles. Due to this fact, geometric modeling of a radar image is clearly simplified as compared to a conventional image as produced by a lens. In the same manner, localization precision will be linked only to satellite position and not to its attitude [1].
[1]
It is necessary to differentiate between three precision types, which in fact, correspond to the three photographic stages:
a priori precision: this is the precision with which we know the satellite's position when it is taking the image (programming precision);
instant precision: at the moment of shooting, the satellite (if it is equipped with navigation instruments) knows its position and can include it in the auxiliary data attached to the image file;
a posteriori precision: a few days after acquisition, this position will be refined by calculations using orbital models (terrestrial potential, friction, etc.) and position measurements previous and subsequent to the acquisition. Then we speak of returned orbits.
It is...
