New Method of Synthetic Aperture Radar Image Reconstruction Using Modified Convolution Back-Projection Algorithm

摘要

Synthetic Aperture Radar (SAR) image reconstruction falls into the class of inverse (deconvolution) problems. A spotlight mode SAR system obtains projections of the ground areas at various look angles. The image of the ground area is then reconstructioned from this set of projections. The Convolution Back-Projection (CBP) algorithm is widely used technique in Computer Aided Tomography (CAT). In this work CBP algorithm has been modified so that it can be applied to image reconstruction from SAR data. A quantitative evaluation using extensive computer simulation of the CBP algorithm for spotlight mode SAR is presented. Its performance is then compared with the FFT method with respect to parameters such as multiplicative noise ratio (MNR), cental processing unit CPU time, and computational complexity. It has been shown that the CBP SAR algorithm is a high quality reconstruction method that can be implemented in real-time. The parallelism inherent in the algorithm can be exploited for parallel architectures that are suitable for VLSI implementation. Samples from each projection can be summed into the proper pixels in the final image, as required by the back-projection operation, so that the final image is produced shortly after the final projection has arrived. (Author)