Research And Application On Iterative Algorithms For Image Reconstruction

Iterative algorithm to reconstruction CT images has advantages because of its simple arithmetic program and widely-using. It may reconstruct images from all kinds of projections which are gotten by various scan modes. But it usually takes a long running time. This disadvantage limits its application in medical and industrial CT. It is well known that it is very important to improve the timing performance. For solving this problem people put forward some methods recently. Owing to compute performance of computer improved, many people pay attention to iterative algorithm again.According to the design in which there are 18 detectors in Hard X-ray Modulation Telescope system of astronomical satellite, the author proposes two ways to fast implementation of the direct demodulation imaging method. One is calculation and storage method for Lucy iteration based on nonzero elements of point spread function of detector collimator, according to the fact that there are a large number of zero elements in point spread function. The other is to substitute Lucy iteration on the ordered subsets of observed data for Lucy iteration on all data in the direct demodulation imaging method. The numerical experiment results show that the imaging convergent speed of the direct demodulation imaging method is accelerated dramatically using the ways proposed in the present paper.Furthermore, the author proposes one way to improve timing performance of parallel interpolative algebraic reconstruction. We make use of relate address of pixels projection and symmetry of image. The numerical experiment results show that the time of reconstruction shortens distinctly.