Based On Block Iterative Fast Algebraic Reconstruction Algorithm

The image of Computed Tomography (CT) is of good quality and highresolution. It can be processed and analyzed digitally Therefore, CT has become anew nondestructive detection technique, which is applied widely in aviation,shipping, automobile, public security and other areas.The filter back-projection (FBP) technique and the algebraic reconstruction.technique (ART) are two of the major image reconstruction methods in CT. Thereconstruction speed of FBP is fast. The image reconstructed by FBP is of highspatial resolution, but with heavy artifacts, while the image reconstructed by ARTis of high-density resolution and is of sligh artifacts. In order to reconstruct image,generally FBP needs complete projection data, but ART is available to eithercomplete or incomplete projection data. The major disadvantage of ART is its lowconvergent speed.In this paper, the author proposes a modified ART (MART) algorithm basedon block-iteration, after analyzing the factors that effect the convergent speed inthe conventional ART. The algorithm converges fast. So it is available toreconstruct moderate and large image by using either parallel or divergentprojection data, as well as the data collected at series arbitrary sampled angles andsampled radii. And it does not require the storage of projection matrix, comparedwith the conventional ART. The numerical results show that MART reconstructsimage with fast speed, high resolution and sligh artifacts.