Study On Airborne Spotlight SAR Imaging And Motion Compensation With High-Resolution

As an active microwave imaging system, synthetic aperture radar (SAR) has found wide applications because of its ability to work at any time of a day and in all weather conditions as well as its range-independent resolution. Also, the low-frequency SAR can perform foliage penetration and ground penetration for detection of concealable targets. Herein, SAR has an important application value in military affairs, such as wide-field surveillance and battle reconnaissance. Currently, it has gradually become a powerful and well-established tool to survey topographic mapping, monitor environmental pollutants and conduct damage assessments. The airborne SAR has great flexibility to form a long aperture and can repeatedly revisit interesting areas with fast imaging. Additionally, the spotlight mode can break the limitations of antenna beam width on the azimuth resolution to obtain high-resolution SAR images.However, due to the presence of atmospheric turbulences and carrier vibrations, airborne SAR raw data are affected by deviations of the platform from an ideal straight flight track, which will significantly degrade the final image quality. In order to restrict the motion error to an acceptable level, effective motion compensation strategy must be involved. With the restrictions to platform load and experimental cost, airborne SAR usually adopts a medium or low precision navigation system and practically cannot fulfill the requirement of high-resolution imaging on the accuracy of wavelength-scale error correction. Hence, autofocus techniques of high precision are employed to estimate phase errors beyond the capability of the navigation system.Based on the fast time domain imaging algorithms and frequency-domain polar format algorithm (PFA), this dissertation studies some key points about the imaging algorithms and motion compensation approaches for airborne spotlight SAR of high-resolution. Besides, the combination between fast SAR imaging and highly accurate motion compensation methods is investigated. The relevant work is supported by the National Nature Science Foundation of China, the Fundamental Research Funds for the Central Universities, and the National High Technology Research and Development Program. The main content of the dissertation is summarized as follows.In Chapter 2, the fast time domain imaging algorithms are represented by fast back-projection algorithm and fast factorized back-projection algorithm; the interpolated-class frequency domain algorithms are defined as original range migration algorithm, its modifications as well as PFA. Basic thoughts, key operations along with advantages and disadvantages are discussed in detail on the basis of the SAR data acquisition geometry and signal model. The connections between the aforementioned algorithms and autofocus techniques are also analyzed. In fact, SAR imaging and motion compensation are essentially taken as a whole. Depending on the available algorithms or their extensions, the latter chapters integrate them with highly precise autofocus processing compatibly, which can further compensate the residual motion errors and aid high quality SAR focusing.Rooting in the basic principle of back-projection integral, the fast time domain imaging algorithms offer a number of advantages over traditional SAR image formations, such as flexible selections of imaging planes, exact compensation of wave-front curvature without geometric distortions. Nevertheless, it is difficult to determine the Fourier transform relationship between the phase history domain and the focused domain, which makes it difficult to integrate motion compensation methods based on autofocus processing with the fast time domain algorithms. Chapter 3 presents an efficient processing scheme combining accelerated time domain algorithm with an accurate autofocus method. The global generalized polar coordinate is firstly introduced, constructing the Fourier transform relationship. The condition for the use of autofocus is also derived. The sub-image fusions in the two-dimensional wavenumber domain are efficiently realized by fast Fourier transform (FFT) and circular shifting. The complete wavenumber spectrum is obtained via seamless connections of the sub-image spectra. Then, inverse FFT is performed in the range dimension and the weighted least square-phase gradient autofocus (WLS-PGA) algorithm is adopted to retrieve residual phase errors. This method only employs FFT and circular shifting to implement sub-image fusions instead of time-consuming, accuracy-restricted two-dimensional interpolation. Meanwhile, it has good compatibility with the high-accuracy autofocus algorithm to estimate the residual motion errors within the radar echoes and to obtain well-focused image.Chapter 4 focuses on a two-dimensional motion compensation approach directly inserted into the efficient PFA for high-resolution spotlight SAR. A coarse error correction is performed by global positioning system/inertial navigation system (GPS/INS) in the range-compressed domain, partly compensating for the envelop error and the phase error. Then, a new envelop compensation strategy, stage-by-stage approach (SSA), is designed and achieves most promising results by removing non-systematic range cell migration (NsRCM) after polar reformatting, which can magnify and change the original form of the motion error. Moreover, a weighted contrast enhancement autofocus algorithm based on spatially variant model is developed to compensate for the residual phase error, remarkably improving the estimation accuracy. By means of image metric of contrast, this scheme is inherently independent of the scene content and does not need bright targets. The experimental results prove its superiority over WLS-PGA on both precision and efficiency.The squint mode has a significant potential to increase the flexibility of spotlight SAR in the sense that it can revisit the particularly interesting area, which cannot be reached instantly by a radar. However, the range-invariant direction of the motion error does not coincide with the ideal flight track. Thus, autofocus algorithms suitable for the broadside mode cannot be directly applied to the squint mode. In order to obtain high-resolution image, longer accumulated time is needed and more serious trajectory deviations are introduced. Chapter 5 describes the center beam approximation and gives the relationship of projecting the three-dimensional motion error into a two-dimensional slant plane. Then, a new coarse motion compensation strategy is discussed with GPS/INS information. Integrated with PFA based on line-of-sight polar interpolation (LOSPI), the azimuth defocusing is in accordance with the direction perpendicular to line-of-sight, which paves the way for autofocus applications. In order to reduce the impacts of LOSPI on residual motion errors, SSA is utilized to mitigate deleterious effects of NsRCM. After the NsRCM correction, the improved total least square estimator-based phase adjustment by contrast enhancement algorithm is presented with a detailed mathematical analysis and applied for spatially variant phase error compensation with high accuracy and efficiency.