<title>Synthetic Aperture Radar Image Bandwidth Compression</title>

Abstract

This paper reports the result of an effort to explore the potential of utilizing the existing compression techniques for the synthetic aperture radar (SAR) imagery. Both adaptive and non-adaptive transform coding techniques were utilized to simulate an end-to-end system, where SAR imagery goes through block-quantization, re-sampling, filtering and encoding, to achieve the desired rate reduction with minimum possible amount of degradation in image quality. Although this investigation utilized limited amounts of SAR data, the approach is not specific to a certain case and is applicable for compression of various SAR imagery. Using simple low-pass filtering, resampling and fast Fourier transform technique, 2:1 or 4:1 data compression leaves all details of the original imagery intact and produces no degradation in image quality based on subjective visual examination. Using semi-adaptive bit-mapping techniques and assigning bit rates of 4, 2, and 1 bit per pixel, further compression on the order of 8:1, with little image degradation, is achievable on most portions of the scene that is examined. This approach has potential for even higher compression ratios if a more adaptive bit-mapping scheme is utilized.