Ship Size and Heading Angle Estimation in Range-Doppler Domain and in Focused SAR/ISAR Images

Abstract:
Ship size estimation is a prerequisite for ship classification. In this paper an eigenvalue decomposition-based ship size estimation method is proposed. The method uses the eigenvalue information of the detected ship pixels and fits an ellipse for estimating the length, width and heading of the ship. The algorithm operates in the range-Doppler domain on coarse resolution image sequences for obtaining rough ship extents as well as on focused high-resolution ISAR and SAR images for more accurate estimations. The proposed method is compared with state-of-the-art algorithms. Airborne and satellite radar data containing real ships are used for evaluating the proposed method.