SAR image based geometrical feature extraction of ships

Geometrical features are widely used in the SAR ship recognition. Two-dimensional (2D) features are sometimes ambiguous and the three-dimensional (3D) ones are hence preferred. Most of the existing researches are based on the 3D coordinate estimation of the scattering centers. These techniques may lose efficiency when working with ocean ships, due to their complicated motion on the waves and the scattering center scintillation among different views. In this work, the ship is simplified as an ellipsoid and the feature extraction problem are transformed into the size and aspect estimation of the ellipsoid. Considering that the projection of the ellipsoid on a SAR imaging plane is an ellipse, whose parameters can be estimated from the ship image, a regularized Least Square (LS) problem is constructed, with SAR images under known elevations and unknown azimuths. The solution of the problem gives the radar azimuth estimation. The 3D geometrical features, such as the size in each dimension, the Length-to-Width Ratio (LWR) and Length-to-Height Ratio (LHR), are evaluated in the end. Experiment with the electromagnetic (EM) simulated images shows the feasibility and efficiency of the ellipsoid approximation and the performances of the estimation are examined via Monte Carlo techniques.