Algebraic Reconstruction Technique Class for Linear Scan CT of Long Object

摘要

The nondestructive testing (NDT) of long object (such as pipeline) has significant value. Computed tomography (CT) is able to generate 3D images and display the inner defects of the inspected object. Sometimes the long object fixed outdoors can't be rotated, and the rotating of the X-ray source and detector is also difficult. So the linear scan CT, translating the X-ray source and detector along longitude of long object synchronously, is considered. However, the projection data of linear scan CT is incomplete because of the limit of the ray-beam flare angle. The algebraic reconstruction technique is able to generate higher quality images compared with the filtered back projection (FBP) method when the projection data is noisy or incomplete. In this paper, simultaneous algebraic reconstruction technique (SART) and Ordered Subsets Expectation Maximization (OSEM) are realized for linear scan CT. Further more, some simulation experiments with different translation step and different flare angle are realized. The results show that the algebraic reconstruction technique class can generate approving CT image in linear scan CT.