MIMO radar imaging and geometrical target properties: a perspective from a classification point of view

Abstract:
Radar systems are able to detect and recognize the general nature of a target, or scattering object, based on information such as its behaviour in space and time. As a non-cooperative target recognition systems, radar systems are able to acquire target recognition information without any cooperation from the target itself by setting up a multidimensional feature space, defined by size, contrast and textural features. Imaging radar systems, such as SAR and Spotlight SAR, are widely used for the classification of potential targets in an region of interest. The paper is addressed to understand the potential and the limits of MIMO radar systems as sensors for acquiring potential target features, particularly the geometrical properties. A crucial step in the creation of MIMO radar images indeed is played by polar interpolation process. Poor interpolators can indeed produce false or spurious targets that are associated with targets that lie within the patch of interest. Moreover they can introduce some distortion which can affect the discrimination of the target dramatically. By comparing 2??D Sinc Algorithm, Bilinear and Sinc-Lagrange interpolation, it has been possible to understand the capability of a MIMO radar of preserving geometrical information of objects of interest. Experiments proved that interpolation process can affect dramatically the estimation of geometrical features. Besides the operating geometrical configuration of the radar is also crucial in order to estimate properly the geometrical features of the objects of interest.