Cross-Parameterization and Compatible Resampling of Point Models

• Main Authors: Hsuan-Yuen Chen, 陳璿元
• Other Authors: Chao-Hung Lin
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/59957608738091983361

碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 96 === In recent years, dense point models scanned from high-precision digital scanner devices have received a growing amount of attention. The point models captured from real-world objects have been extensively studied in many research areas such as computer graphics, scientific visualization, topography, digital archive, and reverse engineering. In this thesis, we propose a novel consistent parameterization and compatible re-sampling approach for 3D point models. The fundamental problem of consistent parameterization is how to consistently embed the input point models into a common parametric domain without any connectivity information. Given a base mesh with only a few vertices, we consistently decompose the input point models into several patches by finding a set of proper paths. We propose a novel approach to trace these paths based on direction fields. Once the consistent patches are obtained, we embed them onto a 2D triangle to obtain the correspondence. In the next process, an efficient approach is proposed to consistently re-sample the corresponding parameterizations. The experimental results show that the proposed approach is robust. Even the points clouds have large different in shape, the proposed approach can handle it well. In addition, we apply the proposed approach to the point morphing and several examples of aesthetically pleasing morphs are demonstrated.