Simulation Of Complex Target's SAR Image And The Application

As a microwave remote sensing means, Sythetic Aperture Radar (SAR) has eminent characteristic of all-sunlight and all-weather, which has wide applications in the fields of military and civilian. Research on SAR image target recognization needs to collect lots of target SAR images with different pose in different imaging condition. It is difficult to get the image set required due to limitation of objective condition. Then simulation of generating SAR image in the given imaging condition and parameter setting is very important.The thesis studies the thory of SAR image simulation and its realization. It consists of geometry modeling of complex targets, RCS computation, generation of SAR simulation image, development of an experimental simulation system and method of target recognization making use of simulation image.The thesis firstly introduces the theory and method of complex target 3D modeling for the application of SAR image simulation, and presents a fast applied modeling method.The thesis secondly mostly states the theory of RCS and the high frequency predict method. Based on it, an improved method of computing RCS, i.e. graphic electromagnetic computing algorithm, is presented., which is integrated with physical optics (PO) method and incremental length diffraction coefficients (ILDC) method.SAR simulation imaging is studied based on the result of RCS in this thesis. SAR imaging theory, imaging mode and imaging algorithm are emphasized, and three simulation methods corresponding to different imaging mode are developed.A SAR image simulation system is set up with integration of relative algorithms. It can get simulation images in different condition by setting many types of parameters, and has a friendly interface and a good maneuverability.The thesis finally does some work on target recognization with simulation images. An integrated approach with target detection and image registration is addressed after the basic theory of target recognization is introduced. This approach can realize target recognization and has better performance than non-integrated approaches.