Low-cost database-free automatic target classification using 3D-ISAR

Automatic Target Classification / Recognition (ATC/R) algorithms based on ISAR 2D images have proven effective but require complex databases. Moving from 2D to 3D systems, can reduce the database size and the costs since target's Computer-Aided Design (CAD) models can be directly used without introducing variations due to the target aspect angle or projection onto the Image Projection Plane (IPP). In this context, we would like to propose a solution that does not need a complex database and that requires only some easily available a priori knowledge about the targets, i.e. target main sizes and generic 2D shapes that can be representative of a target class. This method does not need a training phase and has very low requirements in terms of system memory, system cost and deployment time. This method is therefore useful in situations where there is the need to quickly build an ATC/R system. In this paper, the proposed method has been tested using simulated 3D InISAR reconstructions of airplanes and ships, but the same approach can be easily extended to other target classes because of its easy set-up.