Digital processing of X-band VARAN-S airborne SAR images

A description is given of the techniques developed to process SAR (synthetic aperture radar) data obtained by the CNES X-band VARAN-S airborne SAR during the Agrisar-86 campaign on the Italian test site. The processing was performed starting from the raw sensor data. The main characteristics of the processing are analogical range compression, azimuthal prefiltering, and decimation, all executed onboard in real time. The SAR data were compressed in azimuth by means of a matched filter in the frequency domain. The filter parameters were derived from standard clutterlock and autofocus algorithms. With regard to the autofocus, two partial Doppler bands of azimuth lines at different ranges were used in a correlation process to find the true sequence of the Doppler rate values. Point and extended targets were selected on the resulting full-resolution images to assess the data and processing quality. The radiometric calibration of SAR images was performed utilizing the echo signal from trihedral corner reflectors. The availability of ground truth made it possible to assess the validity of the textural characterization of crop types, where the textural analysis is based on the cooccurrence matrix method.