Research On Fast Image Formation And Frequency Domain Model For The Bistatic SAR

Bistatic synthetic aperture radar(BiSAR) is the system whose receiver and transmitter are placed on different platforms. Compared with conventional monostatic SAR, BiSAR has many advantages, such as long range, strong anti-jamming capability, flexibility, etc. So it has currently been the focus of SAR imaging application. This dissertation mainly focuses fast factorization back projection(FFBP),blocked FFBP and their implementation based on GPU and researches two-dimensional(2-D)spectrum of BiSAR. The main content of this dissertation is summarized as follows.1. Establish the BiSAR’s geometry model and echo model on random flight trajectory; Analyze the slant range history of BiSAR and the resolution of BiSAR by the method of Gradient; Derivating the synthetic aperture time elaborates the feature of azimuth resolution changing in spacel; Introduce some theoretical knowledge about the GPU and describe the programming model based on the CUDA architecture.2. Research the blocked fast BP algorithm. Introduce the procedure of BP and FFBP algorithm; Derivate the relationship between the computation of FFBP and the number of sub-aperture and the quantization formula of the sub-image resolution limit; Since there is a big demand for computer memory of FFBP algorithm, we propose blocked FFBP algorithm. This algorithm block the echo, treated successively and synthesized, which can effectively overcome the drawback of FFBP algorithm. Finally we combine the BFFBP algorithm with GPU parallel processing, verify the effectiveness by BiSAR measured data.3. Research the method of BiSAR 2-D spectrum. Introduce the classic LBF spectrum and derivate the formula of its accuracy limit, illustrate the reason that its accuracy is not high when the squint angle of platforms is big and the velocity of platforms has large differences. Analyze the improved algorithm of LBF spectrum, namely 2D-PSP spectrum. The algorithm introduce the transmitter and receiver platform frequency components, solved by the principal of stationary phase, then obtain the BiSAR 2-D spectrum. At last, we combine the ideas of 2D-PSP spectrum, similarly introduce the transmitter and receiver platform frequency components, use the BiSAR’s geometry model and linear approximation, unite the two linear equations to solve the frequency and then obtain the BiSAR 2-D spectrum. We test and verify this method is effectively able to overcome the shortage of LBF spectrum and its accuracy can match with 2D-PSP spectrum by simulations. Compared with 2D-PSP, the idea of this algorithm is visual used with the BiSAR’s geometry model and the 2-D spectrum is more concise used with linear approximation.