Feature enhanced regularized reconstruction algorithm foe SAR

The biggest problem for automatic target recognition (ATR) is the accurate and efficient extraction of features from SAR images. A new SAR image formation algorithm, which is recognition oriented is developed. The algorithm is based on a tomographic model of SAR observation, using Vector-Entropy optimization approach to reconstruct SAR images from projections. It provides images with higher resolution of scatters and better separate ability of different regions than traditional algorithms. We apply it on synthetic SAR scenes to show the validity of the algorithm.