Wavelet compression of complex SAR imagery using complex- and real-valued wavelets: a comparative study

While many synthetic aperture radar (SAR) applications use only detected imagery, dramatic improvements in resolution and employment of algorithms requiring complex-valued SAR imagery suggest the need for compression of complex data. Here, we investigate the benefits of using complex-valued wavelets on complex SAR imagery in the embedded zerotree wavelet compression algorithm, compared to using real-valued wavelets applied separately to the real and imaginary components. This compression is applied at low ratios (4:1-12:1) for high fidelity output. The complex spatial correlation metric is used to numerically evaluate quality. Numerical results are tabulated and original and decompressed imagery are presented as well as correlation maps to allow visual comparisons.