Analysis-by-Synthesis Compression of Range-Focused SAR Raw Data

We propose a predictive compression scheme for synthetic aperture radar (SAR) raw data based on the analysis-by-synthesis encoding method. The algorithm is inspired by the code excited linear prediction (CELP) algorithm used in speech compression and exploits the correlation across azimuth of the range-focused SAR data. We also extend this approach to include multiple codebooks and obtain the performance of the proposed methods on recorded data. Numerical results show that these algorithms outperform the open-loop predictive quantization schemes.