Tomography Imaging via Triangulating and Neural Computing

摘要

Tomography imaging was originally rooted to the Radon transform, the mathematically reverse analysis presented by Cormack, and the reverse matrix approach by Hounsfield, formed the solid fundament of Computerized Tomography (CT). CT has being paid much attention on its theory, analysis, scanning method, fast algorithm, implementation and applications to medical and industrial areas for fifty years. The problems of the data redundancy and time-consuming of the traditional CT algorithms are paid great interest in. In this paper, we proposed a novel method, the triangulating partition of image and neural computing for tomography imaging. The emphases are paid on the regular triangulating partition instead of the rectangle partition in the traditional CT and the models and algorithms of the neural networks for tomography imaging. The theoretically analyzed and simulated results have shown that the triangulating may reduce data redundancy and the neural computing approach promises a fast and reasonable method to the tomography imaging, especially in some engineering applications such as fluid visualization. The validity and superiority of our approach are examined in this paper.