Clutter reduction for synthetic aperture radar images using adaptive wavelet packet transform

Synthetic aperture radar (SAR) images of ground targets generally consist of target features and clutter from background scattering. In this work, we set out to develop a decluttering algorithm to automatically extract the target image from a SAR image by maximizing the signal/clutter ratio (SCR) using the adaptive wavelet packet transform (AWPT). Our approach is to transform the SAR image to a new domain using the wavelet packet basis. Since a typical target image usually consists of point scatterers and more diffused region features, the multi-scaled wavelet basis is well suited to focus the target image. The clutter image, on the other hand, is statistically uncorrelated from pixel to pixel, and the transformed clutter image under the same set of bases remains unfocused. Therefore, we expect that the SCR can be increased by transforming the original image by an appropriately chosen set of wavelet packet basis. The cost function of our AWPT algorithm is chosen to describe how well the target signal is focused in the transform domain. An efficient basis search algorithm is implemented to find the best wavelet packet basis. Our algorithm is tested using the MSTAR SAR data set.