Comparison on Effects of SAR Data Compression in SQNR and Coherent Change Detection

In SAR applications, the size of data can be quite large, and many data compression algorithms have been developed to reduce the storage requirements for SAR data. Previous research has been done on the compression of both amplitude and raw SAR data. However, the performance metrics used in these compression schemes are often visual-related metrics such as peak-signal-to-noise ratio (PSNR) or signal-to-quantization-noise ratio (SQNR). These metrics have been shown to provide misleading results for data with large dynamic range such as raw SAR data. The present research evaluates the performance of popular compression algorithms for raw SAR data in both SQNR and coherent change detection (CCD), which is superior in detecting subtle, man-made changes in the scene [1]. The results shows that higher SQNR does not lead to better CCD. This finding determines that the popular compression methods currently being used are not optimized for SAR data with respect to CCD.