The Study Of Limited Angle Three-dimensional CT Image Reconstruction Algorithm

Computed Tomography has been widely used in the medical diagnosis and industrial sector for nondestructive detection. In many of the project area, the phenomenon of limited-angle and the limited angle image reconstruction question come out for the projection data because of restrict of test conditions, motif size, motif structure and other factors. There are artifacts and the image resolution is low by using the analytic algorithm and the general iterative algorithm. In the paper, the reconstruction algorithm from limited-angle basic on iteration algorithm is mainly studied. Major to study the contents of the following:(1) This paper gets the following conclusions: by the projection slice theorem limited angle image reconstruction t problem can be converted into band-limited function outside problem. This paper improves an iteration algorithm(GP-SART algorithm)based on the system of the band-limited function extrapolation. This algorithm can reconstruct good image and the edge of image. Imitation experiment and actual experiment have verified the feasibility of the algorithm.(2) This paper systematically introduces the theory of compressed sensing, puts the gradient sparse of CT image into an iterative algorithm, on this basis, this paper put forward a new iteration algorithm (TV-SART algorithm)which is make the image total variation minimum is the best. After every iterative, the image of the total variation has changed by the law of the gradient reduce to improve the image quality. At the same time, this paper had improved the TV₁ gradient, and got the TV_1/2 gradient, it is sparser than the TV₁ gradient to improve image quality and accelerate the speed of a convergent. The reconstruction image of imitation experiment and actual experiment shows that make the image total variation minimum and improve the TV gradient can effectively increasing the speed and precision of the limited angle CT reconstruction and get the good image. (3) Through on these two fundamental research found that the iteration algorithms what used the TV gradient reduce and changed the TV₁ gradient to use the TV_1/2 gradient based on the theory of a limited function extrapolation can effectively to speed up, as well as a good picture on the edge of the experiment. Imitation experiment and actual experiment have verified the feasibility of the algorithm.