Objects recognition with high-Resolution InSAR data and Global Geometric Feature Map

Modern airborne or satellite SAR radar systems provide geometric resolution below ten centimeters. By SAR interferometry from pairs of such images, DEM image can be obtained. Using data of this kind it is possible to build a simple 3D models of remote objects. Similarity measurement between model of unknown object and known database models can be used for classification and recognition. In this paper Global Geometric Feature Map method with improvements is discussed. Since 3D polygonal model can be expressed as a set of facets, the GGFM can fast constitute a spherical transformation to new feature vector containing: normal orientation, area and position of every facet on the model surface. Offline analysis can be done by means of computation of spherical correlation between the GGFM of the object and the models. For online analysis (e.g.: automatic fire-control systems) simplifications of correlation algorithm are required. It can be done by passing the a certain bitmap representation of the data. The experimental results are based on the MSTAR radar images database. In this article we provide the examples of the original GGFM and 2D GGFM analysis.