Airborne Forward-looking Bistatic Sar Motion Parameter Estimation And Parallel Imaging Technology Research

As a special mode of SAR, forward-looking bistatic SAR can get high-resolution imaging of the area under the receiver track, which can’t be imaged by the traditional bistatic SAR, so forward-looking bistatic SAR has its unique advantages. This dissertation studies on the Doppler parameter estimation and parallel implementation of SAR imaging based GPU. The main contents are shown as follows:1. Different models of bistatic SAR are studied, and the characteristics of echo are analyzed. The relationship between the Doppler parameters and motion parameters of plats is discussed. The architecture of GPU and CUDA parallel programming model are studied detailedly.2.A(?)ording to characteristic of Doppler centroid,an estimation algo(?)ithm based on Radon transform is researched; two Doppler frequency rate estimation algorithms based on DCFT transformation and time-frequency rate distribution are studied, and the influence of cross term is analyzed detailedly, the conclusion that dectection factor can accurately estimate the Doppler frequency rate is given; the nonignorable reason of Doppler cubic term is annlyzed, a Doppler cubic term estimation algorithm based on time-frequency rate distribution and Rdon transform is studied.3. As the high-precision forward-looking bistatic SAR simulation and image processing is computationally intensive, the parallel processing of echo simulation and back projection image algorithm is analyzed and implemented, and some program optimization methods which combined with the characteristics of GPU is given. The simulation result verifies the correctness of parallel programming and proves that the GPU’s floating-point computing power is much higher than the CPU, especially for the echo simulation and imaging of large scene. The parallel programming can basically meet the speed of echo simulation and image processing requirements.