Research For Delaunay Triangulation Dividing Based On Spatial Scattered Point And Application In Electromagnetic Scattering

With the rapid development of electronic information technology, the development on electromagnetic scattering analysis of target, especially the researches on electrical large target Radar Cross Section (RCS) and the Inverse Synthetic Aperture Radar (ISAR) imaging become increasingly crucial to accelerate national defense development. Meanwhile, these topics have been current research focuses in Computational Electromagnetics. We found that the decomposition of surface meshes generated in many existing methods are quadrilateral surface elements, however the normal computation is triangle-based, which indicates that the computation of radar targets’ characteristics should rest on triangular meshes.The finite element mesh generation technique has found wide applications in diversity research areas, such as Biomedical Science, Scientific Computing, Computer Graphics, Computer Aided Design (CAD) and Finite Element Analysis (FEA), etc. Being the prerequisite of finite element computing, in which the computation of mesh decomposition has a dominated role and therefore its efficiency and effectiveness greatly affects its followed-up procedures. The triangular mesh decomposition-the oldest mesh decomposition method-has a wide application for its advantages, such as ease of control of the mesh quality, higher efficiency, few human intervention, etc., therefore has been fully developed. Many algorithms have been proposed by researchers, and the most famous one is the Delaunay Triangulation Algorithm. The method has evolved into a crucial direction in many fields. However, there is much room to improve the computing efficiency of many variants of Delaunay Triangulation for scattered point set. It is where our work makes contribution to speed up the computation.This work studies many variants of the Delaunay Triangulation Algorithm, makes improvement on the point-inversion method based Bowyer-Watson algorithm with triangular index, and finds the initial triangle which contains all to-be-inserted points based on random numbers. Based on the Quad-Edge structure, the work implements one triangulation parallel computation method by introduce the popular Map-Reduce parallel programming model. The experimental comparisons among the improved method, the Bowyer-Watson algorithm, and the divide-and-conquer triangulation algorithm over large-scale data indicated that the efficiency of the proposed variant could be greatly enhanced when the quality of the triangular meshes were not negatively affected.Finally, we projected the curved surface of the 3D scattered point set into a 2D space with the projection method, and conducted the triangulation mesh decomposition based on the Quad-Edge structure, then computed its radar scattering characteristics. When compared with the results computed using commercial software, we found that the proposed method had great improvement in efficiency without detriment to the accuracy of radar scattering characteristics.