This thesis presents work on surface reconstruction from imperfect point models. Here "imperfections" refer to the cases such as missing information about point nomials, unknown level of noise, irregular sampling, and/or non-manifold surfaces, resulting in incorrect reconstruction results using existing approaches. Due to the diversity of point acquisition approaches (laser range finders, computer vision methods, etc), imperfections occur often in practice and pose a great challenge to surface reconstruction algorithms. In this thesis, several novel algorithms, targeted towards different imperfections, are proposed. To recover the geometry and the texture of insufficiently sampled regions, we exploit user knowledge about the planarity and the symmetry of man-made objects in indoor scenes and fill the holes in a natural way. After that, there might still be some remaining holes. We then develop a hole filling algorithm as a post processing module to complete the resulting mesh. This algorithm interpolates the interior of a hole based on its vicinity. It can also be used as an independent module for mesh repairing. In addition, we propose a distance-field based algorithm to handle very noisy points without point normals. A global distance field is constructed by propagating the information from interior/exterior regions in a hierarchical and adaptive way. A mesh is extracted later as the zero-set surface of the distance field. Furthermore, a generalized approach is introduced to naturally handle both manifold and non-manifold surfaces, including both non-orientable surfaces and surfaces with boundaries. The input points are first voxelized, enabling easy classification of the local shape for each point. We then locate non-manifold regions (junctions and boundaries) and take special care to mesh them. We also propose to enforce the regularity of the resulting mesh reconstructed from point models. Using a nearly isometric point parameterization, we are capable of controlling the shape of output triangles. The aforementioned algorithms cover a wide range of imperfections and we believe they could handle many problems encountered in practice.
|Advisor:||Kaufman, Arie E.|
|School:||State University of New York at Stony Brook|
|School Location:||United States -- New York|
|Source:||DAI-B 69/01, Dissertation Abstracts International|
|Keywords:||Distance field, Hole filling, Nonmanifold surface, Surface reconstruction|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be