Target Posture Estimation in Synthetic Aperture Radar Images via Attributed Scattering Centers Matching

Target posture estimation is important part in synthetic aperture radar automatic target recognition. In this paper, we proposed a target posture estimation method based on attributed scattering centers matching. Attributed scattering center model perform well in target scattering field description, first, we use attributed scattering center model to predict the strong attributed scattering centers of the target, then approximate maximum likelihood algorithm is hired to extract strong scattering centers from test image and estimate its parameters. A similarity measurement algorithm is designed to evaluate the similarity between model-predicted scattering center set and test image extracted scattering center set. After finding the one-to-one correspondence with highest similarity between two scattering center set, linear regression is hired to estimated posture angle difference between model-predicted image and test image, and target posture angle estimation is updated based on the images posture difference. The accuracy and effectiveness of the proposed method are demonstrated based on experiments using simulated SAR data.