A study on effective numerical analysis of electromagnetics for ISAR image generation
- A study on effective numerical analysis of electromagnetics for ISAR image generation
- Date Issued
- This thesis discusses a study on inverse synthetic aperture radar (ISAR) image generation using shooting and bouncing rays (SBR) and iterative physical optics (IPO). ISAR images show two-dimensional (2-D) spatial distribution of scattering centers of a target, and thus, they are usually used as useful features for radar target recognition. In general, a huge amount of ISAR database of various targets at many aspect angles is necessary for efficient radar target recognition, but it is very expensive and time-consuming to build ISAR database over all targets to be classified and over many orientations. A cost-effective approach to resolve this problem is to exploit a numerical electromagnetic (EM) solver in conjunction with computer aided design (CAD) data of a target in order to generate ISAR images. Well-known numerical methods for EM scattering are the SBR technique and another method is the IPO method. Both methods are useful EM solvers for predicting the radar returns from electrically large target considering higher-order multiple reflections as well as a first-order reflection. In terms of speed, the SBR and IPO are in superior position to the exact solution (moment method), however their process speed is still not fast enough. Furthermore, by an approximate method, they also have a disadvantage of accuracy.
First, we propose a new method to accelerate the monostatic SBR technique suitable for ISAR imaging. The ray-tracing SBR at numerous incident angles demand huge computation time to generate the 2-D ISAR image. Thanks to the bistatic approximation, a significant reduction of the computation time is achieved in the one-shot ISAR technique. However, the bistatic approximation of the monostatic EM solver configuration may cause many spurious artifacts to be shown or some major scattering features of a target to be missed in the obtained ISAR images. For these reasons, to obtain more faithful ISAR images of a target, radar returns based on SBR should be inevitably gathered from monostatic configuration. To achieve this goal, we devise a ray-travel map (RTM) that is a new scheme to store the information of small candidate rays. Therefore, the proposed method can facilitate fast image generation for efficient ISAR database construction. Furthermore, because the proposed method adopts the monostatic approach, it can provide an accurate ISAR image of a target with the reduced distortion.
Second, we propose that the beam tracing SBR is suitable to reduce spurious artifacts of the ISAR image occurred along the cross-range direction. Beam tracing SBR models the incident plane wave by a set of triangular ray tubes whose cross sections are the directly shown partitions of the modeling facets, and each ray tube is traced and recursively subdivided as it reflects from the target. Therefore, this thesis facilitates that beam tracing generates much smaller number of initial rays than the conventional SBR for electrically large target, furthermore retaining the excellent accuracy without the spurious artifacts, because these initial rays are not of frequency-dependent.
Last, we propose the acceleration of the IPO for generating accuracy ISAR image using graphic processing unit (GPU) hardware. ISAR image generated by using SBR technique has the limitation: it shows the inaccuracy image at a cavity. At first, SBR was devised to analyze the scattered field of the cavity, but it did not provide sufficient accuracy in ISAR image. IPO was the newly proposed method for analyzing the scattered field of the cavity with magnetic field integral equation (MFIE). Result of the IPO application for scattered field shows more enhanced accuracy image than that of the SBR. However this method still has a problem with computation speed. To alleviate the computational burden for generating ISAR image, we provide a method which combines adaptive iterative physical optics change rate (AIPO-CR) method and IPO on the GPU.
- Article Type
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.