| Summary: | 碩士 === 中原大學 === 資訊工程研究所 === 93 === This study presents an efficient surface reconstruction algorithm using local vertex characteristics to efficiently generate triangular meshes from unorganized point clouds. The proposed algorithm is fully automatic, efficient, and general. It requires no additional information and user defined parameter. It is an improvement of the efficient surface reconstruction algorithm introduced by Gopi and Krishnan, so the efficiency is inherited. Their approach achieves lower running times at the expense of generality of the point cloud to be reconstructed, so the proposed algorithm is aimed for having weaker assumptions to the point clouds to be reconstructed. Not only does the proposed algorithm correctly reconstruct more general types of point clouds, but it also produces better results. The main contribution of this study includes: 1. Allows for more general types of point clouds to be reconstructed, i.e. it correctly reconstructs point clouds having sharp features, extremely near surfaces, and/or irregular distribution. 2. Generating better results. 3. No user parameter is needed. The above are achieved by two main steps taken: 1. Modification of neighbor points. 2. Propagating triangulation in a smoothness-directed order.
|
|---|
