Using invariants to recognize airplanes in inverse synthetic aperture radar images

We propose a novel method for recognizing airplanes in an inverse synthetic aperture radar (ISAR) image. The method utilizes the conspicuous scatterers at fixed positions in a 2-D ISAR image, and a point-scatterer model is used to model the airplane. The geometric relationships of the scatterers can be used to generate a set of cross-ratio vectors and a database for recognition can thus be built. A pair of stable cross-ratios is applied to the database as a search key to reduce the cost of the search. Furthermore, ordered geometric invariants and the circular matching technique are employed to alleviate the computational complexity. The experimental results show that the proposed method is both robust and effective in recognizing airplanes in ISAR images. (C) 2003 Society of Photo-Optical Instrumentation Engineers.