Study On Improvement And Application Of Algebraic Reconstruction Technique

Computerized Tomography is the most important development in non-destructive detection technique in 20th century. As one of the fundamental algorithm of CT, Algebraic Reconstruction Technique compares with Filter Back Projection, has the advantages like simplicity, efficiency and needless of complete projection data's reconstruction. Especially in the field of Industrial non-destructive detection, Industrial Computerized Tomography is used to detect from sheet metal to huge sophisticate motor. On one hand detection result should has high resolution, on the other hand the difficulty and cost of detection process should be controlled. ART's characteristic is fit for that. However, the drawback influent the traditional ART to be used in industry and clinical studies, which is the huge computation and storage are taked, and the convergence of the reconstruction is very slow.A new field to apply ART is the non-contact measurement of flame temperature field. It is an important technique for factories with large burning furnaces like electricity factory. This technique mainly uses the Charge Coupled Devices with digital-image computing system to reconstruct the three-dimensional temperature field by images of the flame. It has the same principle with CT.This article is divided into 2 parts,1. Discussing the factors of affecting the reconstruction speed and quality of ART. Since the projection coefficient has crucial effect on reconstruction speed and quality. An efficient method based on ray to compute projection coefficient is presented, which classify different situations of grids intercepted by ray. The coefficient matrix can be calculated fast and efficiently as the correlation andsymmetry of the parallel lines. The experiments show that the result of this methodare fast and effective.2. Discussing the appliance of ART in the non-contact measurement of flametemperature field. The nonlinear modeling of radiation transmission is studied andsimplified into linear modeling. Based on the symmetrical structure, simplifiedcomputation of projection coefficient and effective adjustment of projectioniteration order is used to improve ART. The experiment using simulated data showsthat this method is fast and effective.All simulative experiments are built by Matlab. Hope my study may help us doingbetter in industry and medical science.