Application of pixel segmentation to the low rate compression of complex SAR imagery

This paper describes a technique to identify pixels within a subregion ("chip") of a complex or detected SAR image which are to be losslessly compressed while the remainder of the image is subjected to a high compression ratio. This multi-modal compression is required for the intelligent low rate compression of SAR imagery, which addresses the problem of transmitting massive amounts of high resolution complex SAR data from a remote airborne sensor to a ground station for exploitation by an automatic target recognition (ATR) system, in a real time environment, over a narrow bandwidth. The ATR system results might then be presented to an image analyst who, using the contextual information from the SAR image, makes final target determination.