Analysis of SAR images by multidimensional wavelet transform

The description and development are described of a multidimensional continuous wavelet transform for synthetic aperture radar (SAR) image analysis, especially for the analysis of ‘dispersive’ reflectors located on a ground illuminated by a SAR. The usual imaging process makes the assumption that the reflectors are isotropic (i.e. that they behave the same way regardless of the angle from which they are viewed) and white (they have the same properties within the emitted frequency bandwidth). New imaging capacities (offering large bandwidth, low band and large angular excursions of analysis, for example) make these assumptions obsolete. The goal is to present methods based on the multidimensional continuous wavelet transform which highlight these effects for target recognition or identification. Some examples of this analysis are given by processing real data collected in X-band by the RAMSES airborne SAR facilities at ONERA. 1 Description of the problem Radar imaging allows characterisation of the electromagnetic reflection properties of an illuminated target or ground zone. Information contained in the image is often limited to the spatial distribution of the bright scatterers which reflect a part of the emitted radar signal: . for instance, radar imaging is used in an anechoic chamber for stealth analysis to identify the parts of the target reflecting the main part of the emitted signal . in the case of inverse synthetic aperture radar (ISAR) analysis, the goal is to make a spatial image (range and cross-range) of objects (aircraft, satellites) in relative motion with respect to the radar system. This analysis is helpful for recognition and identification purposes . with airborne or spaceborne SAR, the image of the ground is built from the signal collected by a radar in relative motion with respect to the ground. As for ISAR, these analyses allow recognition and identification tasks to be performed on targets and zones of the ground. The radar imaging process is often required to deliver information, which is closely connected to the notion of the ‘optical’ image (infrared image, classical image with CCD system, etc.). Electromagnetic reflection is very different from optical reflection because of the complexity of the electromagnetic wave behaviour in the radar frequency range (HF to W-band). Identification or recognition processes using photo-interpretation are currently under