SAR Image Simulation On Typical Backgrounds And Targets

As one of the most outstanding inventions in the field of radars, Synthetic Aperture Radar(SAR) is widely used in military and civil field because of the advantage of working all time and all weather. Target recognition is one of the important applications of SAR in military field. A large number of SAR images need to be collected for target recognition research. Simulated SAR images contain target structure information. Moreover, they have the advantage of easy acquisition and processing. Thereby, the simulation of SAR images of different incident angles is the basis of target recognition. In this paper, a SAR image simulation software is developed. With this software, SAR images of typical backgrounds and targets are simulated and then the accuracy of simulated SAR images is evaluated. The main work of this paper is as follows:(1) The theory of high frequency electromagnetic scattering is studied. And then a set of SAR image simulation software based on Shooting and Bouncing Rays(SBR) is developed. The input of the SAR image simulation software is 3-dimensional model of the target scene and the output is simulated SAR images of the target scene. SBR adopts both Geometrical Optics(GO) and Physical Optics(PO) which can effectively calculate multiple scattering to calculate target scattering fields. Also, method of equivalent currents(MEC) is combined to calculate the scattering fields of edges. With this software, simulated SAR images of typical backgrounds and targets are obtained.(2) An advanced speckle noise simulation method based on ray tracing is proposed the relationship between target rough surfaces and the distribution of speckle noise of SAR images is derived. The root mean-square height and surface correlation length are two parameters to describe rough surface. According to these parameters, Monte Carlo method is applied to simulate rough surfaces of the target. The height of simulated rough surfaces is applied as random displacements. During ray tracing, random displacements are added at the intersection of each ray pipe and the target surface to simulate rough surfaces of the targets. After that, SBR is applied to calculate the scattering field of the target scene. And then, simulated SAR images with speckle noise are obtained. This method is of high accuracy since it generates speckle noise by simulating the rough surface of the target, which is the fundamental cause of speckle noise.(3) The inversion theory of SAR images for urban structures is studied. In this theory, simulated SAR images of urban structures are inversed into 3-D models. As is introduced above, the input of the SAR image simulation software is 3-D model of the target scene. Comparing the inversed 3-D model of buildings with the input 3-D model of the software, the accuracy of the SAR image simulation software can be evaluated. The estimation method can eliminate the error caused by modeling and focus on the accuracy of the SAR image simulation software.(4) Simulated SAR images of typical military targets and backgrounds such as tanks on rough surfaces, the port and the sea, the airport and runways, aircraft carrier and the sea are analyzed. The simulated SAR images of the tank on the rough surface are compared with SAR images of the tank from MSTAR database to evaluate the accuracy of simulated SAR images. The comparsion result shows that the simulated SAR images of the tank on the rough surface is very precise. For simulated SAR images of the port and the sea, the airport and runways, aircraft carrier and the sea, the qualitative analysis is carried out because of lacking real SAR images of the same scene. The qualitative analysis of simulated SAR images focuses on features of target recognition. The analysis results show that simulated SAR images can well reflect the structure features of targets and thus are qualified to be the database of target recognition.