中文摘要：

根据非透明物体内部不可见的实际，提出了一种基于图片序列的三维表面重建算法。该算法首先利用传统的八叉树算法重建出物体的三维模型，然后利用一种新颖的表面点提取算法提取出物体表面点，最后利用这些表面点进行三角网格剖分，进而重建出光滑的三维物体表面。在表面点的提取过程中，算法对处于不同状态（处于立方体的顶点、棱、面）的点赋予不同的权值，并根据点所占据的顶点、棱以及面的数目对点进行打分，从而根据分值区分内部点和表面点，实现对表面点的提取。实验结果表明，和传统的八叉树重建算法相比，该算法不仅可以极大地减少点的数量，而且可以降低重建误差，使重建的物体表面更加平滑，更有利于表面纹理的粘贴。

英文摘要：

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 triangulation 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.