Compression of SAR raw data via range focusing and trellis coded quantization

Abstract

Recently, there has been a growing interest in the compression of SAR data. Unfortunately, raw data do not exhibit the statistical properties that allow for an efficient compression, and that are present, instead, in the focused data. On the other hand, focusing is too complex to be implemented on board. The authors propose to carry out only the low-complexity range focusing on the satellite, so as to restore part of the correlation and better concentrate the energy in a few samples. To exploit these properties, while limiting the computational complexity, they resort to a variable-rate encoding scheme based on trellis-coded vector quantization. Experiments on real SAR data show a satisfactory performance both in terms of rate-distortion results and of complexity.