On-board medium resolution SAR processing for fast image generation

Abstract

Spaceborne Synthetic Aperture Radar (SAR) instruments are becoming a very important source of information for Earth observation. These systems produce a great amount of data to be stored and extensively processed on ground. However, recent technological developments concerning microprocessors and solid state memories will enable the possibility of on-board SAR data processing. In this scenario, efficient algorithms for generating SAR images are demanded. This new situation requires specially designed algorithms for the processing stages because high-resolution approaches to SAR image generation are still too computationally expensive. We compare and analyse a very efficient SAR processing method for reduced resolution images called spectral analysis (SPECAN) and its variants. This work compares for the first time SPECAN techniques and their variants with other possible candidates. Overall processing evaluation is validated in a simulation environment using ERS-1 raw data. This study compares the computational cost of the algorithms. The conclusion obtained from this study is that SPECAN and especially SPECAN combined with Chirp-Z Transform is a very appropriate candidate for processing SAR data not only in azimuth but also in range. The results show that these new developments are very promising for obtaining fast and simple SAR image generation and they could be suitable candidates for potential on-board implementation.