Study On The Algorithms Of Restoration For The Space Object Image Acquired By Large Binocular Telescope

Obtaining the high resolution images is of great importance in the area of astronomy,remote sensing and space science. Optical interference is a significant technique utilized tomeasure space objects with high angular resolution.As the largest Optical-interferometric system, Large Binocular Telescope (LBT) adoptsadvanced interferometer, LINC-NIRVANA, which is able to achieve instantaneous imagethough Fizeau interferometer and has advanced characters such as common mount，shortbaselines and high angular resolution. LBT, as one of land-based observation systems,is notfree of atmospheric turbulence degradation, although it uses the technique called MCAO（Multi-Conjugate Adaptive Optics）. Compared with equivalent single aperture system, thepartially filled collecting area of LBT system leads to the degradation of image. Imagesacquired by LBT have to be post-processed to obtain high resolution image, which meansimage restoration and reconstruction.In this thesis, the interferometric principles and imaging process of LBT are mainlydiscussed， the theory and algorithms of binocular telescope image restoration are focused on.The four aspects of works we did are listed below:1. The theories of LBT interferometer imaging technology are introduced. Firstly, the stateof the art and observation of LBT are introduced. Secondly, the theories of LBT interferometerimaging technology are reviewed. Thirdly, the current states of LBT arithmetic domestic andaboard again are summarized and its flow and key steps are outlined. Finally, two key techniquesof LBT image restoration are summarized: compensate for the rotation, extraction andextrapolating of PSF（Point Spread Function）.2. The algorithms of LBT image reconstruction are summarized. The model of the LBTimaging is analyzed and the single frame iterative restoration model can be obtained by SGM(Split Gradient Method) algorithm minimizing the cost function. The single frame iterativemethods are introduced to the multiple frame iterative method. Some forms are given to someSGM decompose of the cost function of Gaussian and Poisson noise statistics and some usualregularization terms. The special approach, Richard-Lucy algorithm, is deduced from the generalapproach. The experiment results of the simulated LBT data show the effectiveness of thealgorithm.3. Multi-frame iterative blind deconvolution algorithm based on Fourier domain constraintsis proposed. A new kind of prior knowledge based on the difference of image data informationboth in space domain and Fourier domain is proposed, the Fourier spectrum are analyzed and the limitation in Fourier domain of PSF and the target image are given. It is proved that thismethod can realize image blind deconvolution, get better details of reconstruction image, andhigher fidelity image.4. Multi-frame blind deconvolution algorithm based on the Poisson-Gaussian mixed noise ispresented. In accordance with the practical situation, the noise model based on Poisson-Gaussianmixed distribution is introduced. According to the center-limiting theory, noise model iseffectively approximated, and then logical regularization terms are built to improve theconvergence of the algorithm. Experiments with simulating and observing data proved that thealgorithm has good performance and produce better restoration image.