Improvement of X-Ray CT Image Quality by Use of Prior Information

摘要

Most objects processed by industrial CT contain ma- terials having a large ray-absorption coefficient (e.g., met- al). In such cases, streak artifacts (SA) occur due to excessive X-ray attenuation. If a region having a large absorption coefficient is in a section to be examined, this affects the whole measurement. However, since the size, shape, material, and structure of an industrial object are known from its design information, this region can be heated separately. This paper describes an algorithm to reconstruct such a special region by using a sequential approximation. Since previously known ray-absorption co- efficients are given to the rest of the region in the object, the influence of SA can be reduced. Computer simulations with several phantoms were carried out, and the proposed method was demonstrated to have a very significant effect. If the abnormal part is large, the effect of the method is degraded. However, it has been found that this degradation is suppressed by increasing the directions of ray projec- tions. If regions of high absorption and low absorption are in a single region, the effect of image improving by the method is also degraded, but a solution of this problem has been found.