Study On Sparse Signal Processing Of High-resolution Radar Imaging

As an active tool of microwave remote sensing, the radar imaging technology is characterized by all-weather, all-day, long-distance work, high-resolution and etc. Herein, synthetic aperture radar(SAR) has played an important role in battlefield reconnaissance, terrain mapping, resource exploration, environment survey, and so on. Meanwhile, inverse synthetic aperture radar(ISAR) is able to image the targets in airspace, aerospace and sea. And both of them can significantly enhance the capability of information acquisition in remote sensing.The current trend of SAR/ISAR development is wide-swath and high-resolution imaging. With the increasing of requirements, SAR/ISAR is gradually developed to multi-functions/modes and networked sensors from single ones. Meanwhile, the manner of data acquisition now behaves as diversity, where it is changed from single channel, polarization and view to the multiple ones. In the case of multi-dimensional measurements, current radar imaging techniqes suffer from several significant problems, i.e. huge amount data storage, sparse aperture imaging and parameter /information extraction.Focusing on several bottleneck problems of current radar imaging, this dissertation studies the techinqiues of high-resolution radar imaging in a sparse way, which aim to resolve the huge data amounts in wide-swath imaging, sparse aperture imaging and parameter/information extraction of the targets. The relevant work is supported by National Basic Research Program of China(973 Program, No. 2010CB731903), National Science Foundation of China(No. 60890072 and No. 61301280) and others. In this dissertation, we put our efforts in non-ambiguous imaging of sparse sampling data and imaging joint with information extraction.The main content of this dissertation can be summarized as follows:(1) Robust autofocusing approaches of ISAR imagingIn the first part, we propose a sparse representation approach of ISAR imaging joint with phase error correction, effectively improving the imaging performance in low signal-to-noise ratio(SNR). Based on statistical modeling, the sparse representation ased ISAR imaging is created to treat phase errors as model errors. Then, a modified quasi-Newton method is presented to realize non-ambiguous azimuth imaging joint with phase error correction. Herein, the fast Fourier transform(FFT) and matrix Hardmard multiplication operations are applied to promote the efficiency of solution. Finally, the sparse representation approach is extended to sparse aperture imaging. Incorporating with conventional autofocusing techniques, the modified sparse representation approach can effectively overcome both the sparse aperture and phase errors to get a high-quality ISAR image.(2) ISAR imaging joint with azimuth scaling The second part studies sparse aperture ISAR imaging and scaling. For uniformly rotated targets, a joint processing approach of phase adjustment, migration through resolution cells(MTRC) correction and sparse aperture imaging is presented based on parametric sparse representation, which can effectively overcome the difficulties of phase error correction and non-ambiguous image reconstruction in high-resolution imaging. For maneuvering targets, the sparse representation is applied on sparse aperture data to realize scattering center extraction, chirp parameter estimation and Range-Doppler imaging. Then, by exploiting the diversity of the chirp parameters for multiple scattering centers, they are used to estimate the rotational motion of the targets and the azimuth scaling can be achieved. Finally, we extend the imaging approach of maneuvering targets to consider the presence of MTRC in ISAR image, where the imaging model and solution are modified. In comparison, the imaging performance in the cases of low SNR and data sampling amounts can be effectively enhanced.(3) Joint multi-channel 3-D imaging and rotational parameter estimation in In ISAR In the third part, the sparse aperture 2-D imaging approach of maneuvering targets is extended to the case of joint In ISAR multi-channel 3-D imaging. Here, we present a joint multi-channel sparse representation approach. In comparison with single channel approach, the proposed joint multi-channel approach can effectively improve the performances of scattering center extraction, parameter estimation and 2-D imaging. To reduce the errors in scattering center extraction and parameter estimation, target 3-D geometry reconstruction joint with rotational motion estimation is executed using an optimization method in an iterative manner. As a result, the outliers and errors in samples can be effectively removed to realize accurate 3-D geometry reconstruction and rotational parameter estimation.(4) Wide-swath imaging of sea multi-ships The fourth part focuses on single-channel wide-swath SAR imaging of multi-ship targets. In overall consideration of system designs, the radar beam is controlled to scan along range direction to realize a wide swath observation, where each swath corresponds to sparse aperture measurements. According to the principle of Range and Doppler decoupling, the dictionary of echo modulator is created and non-ambiguous imaging approach is presented to realize wide-swath and high-resolution imaging. In addition, the performance of echo modulator is analyzed to ensure a high probability of successful imaging.(5) In SAR image noise reductionIn the last part, we present a sparse approach of amplitude and interferometric phase for In SAR image based on Bayesian representation. The In SAR image formation is treated as joint sparse regularization of amplitude and interferometric phase, where a sparse prior of In SAR image is exploited. In this way, simultaneous de-speckling and interferometric phase noise reduction can be realized to improve the noise reduction performance. Then, we extend the signal-channel approach to joint In SAR multi-channel one to further protect the cross-channel information.