Reconstruction Of Radio Astronomy Image Based On Compression Sensing

The radio emission of the celestial body can reflect its own features and state, so radio observations have been a very important method in the astronomical research. While the aperture synthesis radio telescope is an important part of the radio astronomy instruments, the research on radio images has a great practical value. In general, correlative interferometers in a fixed baseline on earth can get a Fourier component of the celestial body. If we know all the Fourier components of an image, the image can be reconstructed accurately. However, because of the limited numbers of antenna, Earth’s rotation and other reasons, the data obtained is incomplete and sub-sampling, and thus to get a more exact solution on this data is the most potential way. Currently, most of the radio astronomy images restoration methods are based on the CLEAN algorithm, the maximum entropy algorithm and their improvements. As classic image restoration methods in radio astronomy, they do not completely solve the under-sampling data reconstruction problem fundamentally. Therefore, this thesis focuses on the radio astronomy image restoration. Compressed sensing (CS) theory first appeared in the magnetic resonance image restoration from incomplete Fourier samples, and it is very similar to the physical model of the radio images restoration. Therefore, combining with the aperture synthesis radio observational techniques, and the CS theory, this thesis is focus on solving the reconstruction problem from the sup-sampling image data. The main contents are as follows:(1) Firstly, this thesis introduces the background and significance of the radio astronomy research based on CS theory, and describes the current research methods about the aperture synthesis radio observational techniques and CS theory. Then CS concept, principles and implementation has been described, including three key components:the sparse representation, measurement matrix and reconstruction algorithm. At last the prospects for compressed sensing signal processing is discussed.(2) Next, a in-depth study on the radio astronomy image reconstruction methods based on CS is proposed. First the thesis introduces two categories reconstruction methods, namely the greedy method and convex optimization method, and then describes the concept of the principle of the two methods and related algorithms. Based on the influence of the reconstruction methods on the reconstruction image quality, this thesis conducts an experimental analysis of each reconstruction algorithm, and an improved compressive sampling based on greedy method is proposed at last.Considering the characteristics of celestial brightness distribution, On the base of Fourier sparse components, aperture synthesis radio observational techniques and compressed sensing radio astronomical image reconstruction method, we propose a new aperture synthesis radio observational and compressed sensing radio astronomical image reconstruction framework. Then we compare the experiment result with traditional image reconstruction method. Our method show high efficiency in the complex field, and have better image quality than other methods.In addition, the thesis also discussed the possibilities of using compressed sensing radio in next generation of synthetic aperture radio telescope. It has a certain value for telescope image reconstruction technologies.