A Study on Digital Image Reconstruction of SAR Data

Abstract

This paper firstly describes the principles of SAR(Synthetic Aperture Radar) imaging, i.e. resolution concept, frequency pulse modulation, pulse compression and Doppler effects. In the real aperture radar systems, the azimuth resolution depends on slant-range, wave length and antenna length, therefore it is greatly limited by the sensor itself. On the contrary SAR can achieve much higher resolution by storing observed signals and processing afterward.
Secondly the analytical expression of transmitted and received signals is studied, and the digital processing of them to reconstruct SAR images is also theretically treated in detail, in terms of both range and azimuth directions, as one of compression problems of two-dimensional impulse response which is reflected from a point target. In this paper, matched filters in frequency domain are described along with some detailed evaluation of them. This processing is mainly consists of 4 steps, (1) holographic data generation, (2) real-complex data conversion, (3) range compression and (4) azimuth compression. In addition to the azimuth compression, multilook images are stacked to reduce the speckle noises which appear in reconstructed images because of the coherency of microwaves used.
As an actual example, a two-dimensional impulse response generated by computer with SEASATSAR parameters was successfully compressed to a sharp single impulse.
Finally a SEASAT-SAR image around LOSANGELS is reconstructed to show the validity of the digital processing presented by the authors.