「三次元NMR画像からの三次元物体形状の自動抽出法の研究」

要旨

Abstract-A method to extract shapes of bubble polyhedra from 3D NMR image data of foams has been developed.The method consists of four steps : (1) computation of series of maximum intensity projection (MIP)images from3D image data to visualize the network structure of bubble edges, (2) interactive measurements of 3D-coordinatesof vertices of bubble polyhedra on X-window using MIP images to extract polyhedral edges, (3) building up polyhedral faces from the edges data, and (4) building up polyhedra from face data. The building methods are based on selection of polygons and polyhedra from enormous number of combinations of edges and faces under some selection rules. Since the computation time for the selection increased exponentially with the number of the input data, a“sweeping”method ( this is a kind of “divide-and-conquer”）in which a cubic volume region containing a bubblepolyedron is swept in the 3D image area has been developed to reduce the computation time. By using this method,we have succeed in automatically building eight bubble polyhedra from the edges data less than 45 minutes.