Investigation On Imaging Algorithm For Correlated Imaging System Based On Compressed Sensing

Correlated imaging has attracted much attention from researchers from home and abroad. In correlated imaging, two or more sensors are utilized to measure optical field coincidentally. The information of an object is then obtained based on the second-order or even higher order correlation of the optical field. An imaging system based on correlated imaging can be used to obtain image of ultrahigh resolution by breaking through the diffraction limit. It is also more tolerant of environment factors such as cloud, frog and etc. Therefore, correlated imaging system has a promising future in both civil and military application. However, its traditional correlation algorithm demands many times of detection of an object, which results in the slow imaging speed of the imaging system, and the tremendous expense on data transmission and storage. Consequently, the development of correlated imaging research will be constrained.Currently, research on correlated imaging is mostly conducted centering on three aspects: the theoretical model, the imaging system and the imaging algorithm, among which the last one is of significant importance. By using an appropriate imaging algorithm, the image of an object can be obtained with lesser measurements and better performance. Thus, both the time of imaging and the amount of data is reduced, thus promoting the development of research and application of correlated imaging.This dissertation is focused on the research of the correlated imaging algorithm based on compressed sensing in order to lessen the number of measurements and save data space. First, the theory of compressed sensing is introduced over three aspects, the sparse presentation of images, random projecting matrices, and the reconstructing function, with a focus on the algorithm of the reconstructing function. Simulation of three different algorithms is conducted to choose a most practical one. In the compressed correlated imaging algorithm, the projecting matrix is constituted of real experiment data, thereby presenting certain physical meaning and influenced by system parameters and background noise. The performance of the imaging result is therefore degraded. So, an improving proposal is raised based on a simple yet effective processing upon the projecting matrix, which improves the imaging performance by reducing the impact of system parameters and background noise. Simulation and experimental results indicate that imaging results of better performance and clearer outline are obtained using the advanced algorithm proposed. More surprisingly, it is also proved by experiment results that the image of an object can be obtained with an even lower sampling ratio by using the proposed algorithm. Therefore, the proposed algorithm can be employed to fasten imaging speed, reduce the cost of data processing and storage.