Three-Dimensional Surface Reconstruction Based on Image Sequence
Yuan Miao,Jiajun Wang,Shoukuai Liu
DOI: https://doi.org/10.3969/j.issn.1004-9037.2011.04.006
2011-01-01
Abstract:A three-dimensional surface reconstruction algorithm based on image sequence is proposed according to the fact that the internal part of an opaque object is invisible. Firstly, the traditional octree reconstruction algorithm is employed to reconstruct the three-dimensional model. Secondly, a novel algorithm is proposed to extract the surface points from the results of the octree reconstruction. Finally, the surface points are subdivided by the Delaunay triangula-tion algorithm, where the smooth surface is obtained. During the process of surface point extraction, different weights are assigned to different points according to their states (in the vertex, the edge, or the face of a square), and each point is scored according to numbers of vertices, edges and faces it occupies. According to the scores corresponding to the points, they are classified into two classes, that is, the internal points and the surface points. Hence, the latter can be extracted. Experimental results show that the proposed algorithm outperforms the traditional octree algorithm not only in a significant reduction of the number of points involved, but also in a decrease of the reconstruction error, thus achieving a more smoothing surface, which is of critically important for texture mapping.