Doppler Ambiguity Resolving For SAR And Processing Algorithms For Bistatic SAR

Synthetic Aperture Radar (SAR) is a very important remote sensor. As compared with optical sensors and other microwave remote sensors, SAR has several unique advantages. This dissertation addresses issues of SAR imaging. The work of this dissertation mainly focuses on two aspects:The first one is a thorough study of resolving Doppler ambiguity which is an essential procedure for SAR Doppler centroid estimation, and the second one is a detailed investigation of digital processing algorithms for bistatic SAR data.The main content of this dissertation is summarized as follows.●The first part of this dissertation presents a novel method for resolving Doppler ambiguity. By analyzing the echoed SAR signal, it is found that in the compressed azimuth time and range frequency domain, all targets span the same range frequency bandwidth and exhibit the same slope which is just proportional to the Doppler ambiguity number. The aforementioned fact just makes the basis for the proposed method of Doppler ambiguity resolving. To measure the above-mentioned slope, a simplified Radon transform is utilized. The use of entropy in finding the maximum concentration of the Radon transformed image can improve the robustness of the method. The proposed method directly gives a reliable estimate of Doppler ambiguity number, and is independent of the baseband Doppler centroid estimation. Simulation and real SAR experimental results show that the proposed method works well in medium to high contrast scenes. Besides, the proposed method has another advantage that slowly-moving targets have no influence on the estimate of the Doppler ambiguity number, as long as the Doppler shift induced by these slowly-moving targets is not greater than one PRF.●Based on Neo's method-of-series-reversion (MSR) 2-D spectrum, the second part of this dissertation proposes two OMEGA-K algorithms for focusing the bistatic data:The first OMEGA-K algorithm is an easily-implemented algorithm and the second one is an improved algorithm. The key step of deriving the two OMEGA-K algorithms is to derive their range frequency mapping function, and the key of deriving the range frequency mapping function is how to linearize the 2-D spectrum. For the easily-implemented OMEGA-K algorithm, the 2-D spectrum is linearized by adopting an angle approximation, whereas for the improved OMEGA-K algorithm, the 2-D spectrum is linearized by introducing a reference vector on the ground plane of the bistatic geometry. For the improved OMEGA-K algorithm, the optimum direction of the reference vector, along which the best performance of OMEGA-K algorithm is achieved, is also determined; this optimum direction is approximately such one along which all targets have the same instantaneous Doppler frequency as the reference target at zero azimuth time. Simulation results show that for moderate bistatic configurations (i.e., the bistatic baseline is relatively short), both the two algorithms can yield a well focused SAR image (in this case, the first OMEGA-K algorithm has a low computational load and thus is more efficient), whereas for extreme bistatic configurations (i.e., the bistatic baseline is relatively long and/or the synthetic aperture time is relatively long), the second OMEGA-K algorithm can achieve a remarkable performance improvement over the first OMEGA-K algorithm, and thus for more extreme configurations, the second OMEGA-K algorithm (i.e., the improved algorithm) is recommended to be used.●The third part of this dissertation gives an analytical method to update the range derivatives for the MSR-based range Doppler algorithm (RDA). The proposed method mainly exploits two points:The first one is the introduction of a reference vector on the ground plane of the bistatic geometry, and the second one is the use of the principle of series reversion. The proposed method is accurate and easy to implement, since it is analytical. Note that the proposed method of updating the required parameters along range is also applied to other bistatic processing algorithms (see the fourth and sixth parts of this dissertation which will be mentioned in the following).●Based on Neo's MSR 2-D spectrum, the fourth part of this dissertation derives an extended MSR (EMSR) 2-D spectrum for bistatic data processing. The EMSR 2-D spectrum is suitable for handling a special bistatic configuration, called the hybrid spaceborne/airborne bistatic SAR configuration. The key step of the derivation of the EMSR spectrum is to derive an equivalent translational invariance azimuth time (TIAT) for the bistatic geometry. It is found that the equivalent TIAT of the bistatic geometry can be physically interpreted to be a weighted sum of the transmitter TIAT and the receiver TIAT, and the weighting factors are just the ratios of the azimuth FM rate of the transmitter platform and the receiver platform to the overall azimuth FM rate, respectively. In the EMSR spectrum, an important parameter called the equivalent platform velocity, which is used to register the azimuth coordinate of the focused SAR image to the ground coordinate, is also derived. Then, we apply this newly derived EMSR spectrum to the MSR-based RDA. In developing the modified RDA, the new method for updating the required range derivatives for the MSR-based RDA which is given in the third part of this dissertation, is adapted to the newly derived EMSR spectrum. Besides, in this part, we also investigate several issues related to the modified RDA, including the SAR image registration and the determination of the azimuth-invariance region size.●Based on the way of transforming the bistatic system into an equivalent one, the fifth part of this dissertation derives a new 2-D spectrum for bistatic SAR processing. For the newly derived 2-D spectrum, two phase terms are identified, the first one being an equivalent monostatic (EM) contribution and the second one corresponding to a bistatic deformation (BD) contribution. The key step of formulation of the 2-D spectrum is to compensate the difference between the double square-root (DSR) range equation and an equivalent single square-root (SSR) range equation by using an algebraic operator which consists of a third-order term plus a fourth-order term, so that the bistatic system can be transformed into an equivalent monostatic one. The new 2-D spectrum can be regarded as an extension of Sun's spectrum, and it also incorporates the concept of Rocca's smile operator. The new 2-D spectrum is exactly accurate up to the fourth-order phase term and it can handle a general bistatic SAR configuration with nonparallel flight tracks, unequal radar velocities and different heights of the transmitter and receiver. Besides, in this part, we also compare four bistatic 2-D spectra, i.e., the spectrum derived in this part, Sun's spectrum, LBF spectrum and the extended LBF (ELBF) spectrum. Theoretical analyses and simulation results show that the spectrum proposed in this part applies to a more general bistatic configuration and can achieve a better focusing performance than the other three spectra.●Based on the new bistatic 2-D spectrum derived in the fifth part of this dissertation, a bistatic RDA is developed in the sixth part of this dissertation. In developing the RDA, the secondary-range-compression (SRC) term was first factored out from the EM term and then was incorporated into the BD term so that a new BD term is formed. To handle the problem of the space dependence of the new BD term, we can divide the whole scene into segments and then remove the new BD term using phase multiplies blockwise in the 2-D frequency domain, due to its slowly-varying property in space. To address the problem of the range dependence of the RCM term and the azimuth modulation term, the new method which is proposed for the MSR-based RDA in the third part of this dissertation is adapted, so that after adaptation, this method can effectively update the three equivalent parameters (i.e., the equivalent range, the equivalent velocity and the equivalent squint angle) along range. The proposed RDA is efficient and easy to implement, since the expressions of the RCM term and azimuth modulation both have a concise and closed form (rather than a Taylor expansion version). The proposed RDA almost has the same processing steps as the conventional monostatic RDA except for an extra procedure of updating the required equivalent parameters, which means that the conventional monostatic RDA can be readily applied to process the bistatic SAR data with minor modification. Besides, in this part, based on the conventional two-step method, we also present a motion compensation (MOCO) method for the proposed RDA. The key step of the proposed MOCO method is to perform range-dependent MOCO along the reference vector. Simulation results show that the proposed MOCO method is effective.