The Study Of Cluster-based Parallel Algorithm Of Image Reconstruction For Synthetic Aperture Microwave Radiometer

Synthetic aperture microwave radiometer uses an antenna array which is formed from a group of small antennas to receive and interferometric techniques is adopted to passively imaging. Due to the solution of the problem of antenna size, it can obtain the higher space resolution and has a good application prospect. The image reconstruction algorithm for synthetic aperture microwave radiometer include two class: Fourier method and G-matrix method. The image acquired through Fourier method is not expected, and the image acquired through G-matrix method is what we want. However, the theory and algorithm of G-matrix method are complex, which become the reson of out of memory and long time-consuming. In order to obtain clear image and unfreeze the limitation of memory and time consuming, this paper put forward that using parallel computing technology to study the parallel algorithms of synthetic aperture microwave radiometer imaging.This article firstly introduces the development of both synthetic aperture microwave radiometer and the image reconstruction of synthetic aperture microwave radiometer and parallel computation. Then a brief introduction to the basic principle of synthetic aperture microwave radiometer imaging and the algorithm of image reconstruction is arranged in this paper. The algorithm complexity of synthetic aperture microwave radiometer image reconstruction is analyzed from two aspects of theory and experiment. On this basis, we excavate deeply the parallel features of synthetic aperture microwave radiometer image reconstruction algorithm and design task allocation scheme of parallel algorithm and concrete realization of the parallel algorithm. In the end, the performance testing of the parallel algorithm is done on the cluster platform. The results prove the validity of the parallel algorithm, and the limitation of memory is broken through, and the time-consuming is shorten, which effectively solve the problem of vast data and long time-consuming, and has a good application prospect.