Study On Imaging Algorithms For Bistatic SAR

Bistatic synthetic aperture radar (BiSAR) is the SAR systerm whose transmitter andreceiver are placed on different platforms. This configuration increases the flexibility in dataacquisition and then expands the data usability, and has been implemented in recent years.Due to the presence of the addition of two square–root terms in the range equation from theBiSAR target, one cannot analytically derive an accurate transfer function in the2–Dfrequency or range Doppler domain using the principle of stationary phase (POSP) that iswidely used in the processing of the MoSAR data. Thus, algorithms that are capableprocessing the MoSAR data are not readily to handle the BiSAR data. This dissertationmainly focuses on two aspects:(1) the method to obtain the2–D spectrum of the BiSAR;(2)the algorithms for BiSAR data focusing. The main content of this dissertation is summarizedas follows.1. A novel method to obtain the stationary phase for general BiSAR is developed in thefirst part. Actually, the stationary phase is the implicit function of the azimuth frequency (AF).It can be expressed as a series of AF. Using derivatives of an implicit function (DIF), thecoefficients of each term can be achieved, as well as the stationary phase. The final2–Dspectrum is able to obtain. Similar to the method of series revision (MSR), the accuracy of the2–D spectrum can be controlled by keeping enough terms. However, the DIF is thegeneralization of the spectrum obtained by the MSR. Also, using the DIF method one canaccurately derive the2–D spectrum for a generally configured BiSAR, and simplify theexpressions of the stationary phase and round–trip range history for the BiSAR under anazimuth–invariant configuration. The replacement of two square–root terms with one term isespecially advantageous in implementation because changes to an imaging processingalgorithm are minimal. As an example, we modify the RDA using the DIF–derived spectrumto process azimuth–invariant BiSAR data. Therefore, a novel method to process BiSAR rawdata has been developed.2. In the second part, an equivalence–based2–D spectrum for general BiSAR isproposed. First, a monostatic equivalent range equation that consists of a monostaticequivalent component and a BiSAR compensation component was proposed. There were fivemonostatic equivalent parameters (MEPs) in both components. We constructed a newhyperbola of one square–root term to replace the addition of two square–root terms in thegeneral BiSAR range equation. The hyperbola consisted of an equivalent monostaticcomponent with three MEPs and a bistatic compensation component with two MEPs. Using the MEPs, we approximated the BiSAR range equation in the form similar to that of amonostatic SAR (MoSAR). Then, with the principle of stationary phase (POSP) and rangeequation, the2–D spectrum of a point target was analytically and directly derived. Themonostatic equivalent range equation is more accurate than that of a polynomial up to the4thorder. Hence, the obtained2–D spectrum is concise with very high accuracy even under someextreme conditions, such as super high resolution and very high squint angles.3. The equivalence–based spectrum was next implemented into the range Doppleralgorithm (RDA) that processes the azimuth–invariant BiSAR data. The modified RDA hasthe same processing steps to the conventional one. However, different from the conventionalRDA for monostatic SAR, the range cell migration correction (RCMC) kernel and the phaseof azimuth compression (AC) filter are all functions of the five space–variant MEPs.Consequently, conventional RDA is not able to handle the azimuth–invariant BiSAR datawhen the five MEPs are unknown. On the basis of an equal Doppler center line (EDCL), aninnovative but a simple approach was developed to solve the five MEPs. The proposedmethod is accurate. It is much easier than the numerical method to implement. Under a TIconfiguration, satisfactory results have been simulated. Promising outcomes were alsoobtained in the analysis of acquired BiSAR data.4. In the fourth part, a novel method to obtain the formulations of the return signals inthe two–dimensional (2–D) frequency domain for both monostatic and bistatic SAR isconducted. In this part, the modified effective wavelength (MEW) is firstly used, so that the2–D spectrums can be derived directly from their imaging geometries. For MoSAR, the2–Dspectrum is obtained without a lengthy derivation by using the widely–used principle ofstationary phase. For the BiSAR, based on the assumption that the azimuth time durations ofthe transmitter and the receiver are the same, two individual MEWs can be derived, as well asthe2–D spectrums that are both concise and of high accuracy. Similar to MSR, the accuracyof the2–D spectrum for BiSAR can also be controlled by keeping enough terms. This new2–D spectrum can be used to data focusing even under some extreme cases (such as very highresolution and very high squint angles).5. Later, two modified omega–K algorithms based on the two2–D spectrums aredeveloped to process MoSAR and translational–invariant BiSAR data. Furthermore, asimportant processing steps of the proposed omega–K algorithms, a modified referencefunction multiplication and a modified Stolt mapping, which are much more suitable for SARdata processing than the conventional ones, are proposed. Simulations under a wide range ofMoSAR and BiSAR systems parameters are conducted. Finally, the proposed algorithms areapplied to the analysis of acquired data and the results confirm not only the validity of the derived2–D spectrums for both MoSAR and BiSAR but also the effectiveness of theproposed method.