Recognition-oriented Bayesian SAR imaging

We develop a recognition-oriented SAR image formation framework from a Bayesian perspective. In our framework, the prior knowledge of the targets and scene will be treated as restriction to the expected image based on Bayesian inference within an inverse problem frame. As a result, an image beneficial for recognition with background suppression and the important feature enhancement can be achieved. Results on both simulated and MSTAR data validate the performance of our proposed approach.