SAR image compression at very low bit-rate

SAR (synthesis aperture radar) generates huge amounts of data which may easily exhaust the available capacity of downlink or storage device of the spaceborne SAR system. So there are special interests in SAR image compression. Common methods suffer from "ringing" effect at very low bit-rate when using wavelet codec: as a result poor quality is achieved. Our approach uses a Gaussian filter as a low-pass filter, which not only acts as a despeckle tool, but also makes the image signal smooth enough so that it can be compressed at very low bit-rate with little ring bell artifacts. Then the image can be encoded with familiar wavelet codec such as JPEG2000. At the decode side, we apply a Wiener filter to act as a post-process which restore important edge information. The experiment results present on RADARSAT images show good visual qualify at 0.25 bpp.