Relative error-constrained compression for Synthetic Aperture Radar data

Near-lossless compression, i.e., yielding strictly bounded reconstruction error, is extended to preserve the radiometric resolution of data produced by coherent imaging systems, like Synthetic Aperture Radar (SAR). First a causal spatial DPCM based on a fuzzy matching-pursuit (FMP) prediction is adjusted to yield a relative-error bounded compression by applying a logarithmic quantization to the ratio of original to predicted pixel values. Then, a noncausal DPCM is achieved based on the Rational Laplacian Pyramid (RLP), recently introduced by the authors for despeckling. The baseband icon of the RLP is (causally) DPCM encoded, the intermediate layers are uniformly quantized, and the bottom layer is logarithmically quantized. As a consequence, the relative error, i.e., pixel ratio of original to decoded image, can be strictly bounded around unity by the quantization step size of the bottom layer of the RLP. Experimental results reported on true SAR data from NASA/JPL AIRSAR show that virtually lossless images can be achieved with compression ratios larger than three.