| Summary: | 博士 === 國立成功大學 === 資訊工程學系碩博士班 === 97 === Texture mapping is a common and very useful technique in computer graphics. Texture mapping adds the detailed information of the user-specified image to the 3D models and represents the rendering result more vividly and realistically. This thesis contains three efficient algorithms from texture mapping on 3D surface without positional constraints to texture mapping with hard positional constraints.
The first one is to achieve distortion-free texture mapping on arbitrary 3D surfaces without constraints. To texture 3D models, we propose a scheme to flatten 3D surfaces into a 2D parametric domain. Our method does not require the two-dimensional boundary of flattened surfaces to be stationary. The first parameterization scheme can be efficiently realized by a linear sparse matrix system and yields interactive performance.
The second and third algorithms are texture mapping with hard positional constraints. Texture mapping with positional constraints is an important and challenging problem in computer graphics. The idea of second one is to partition the parametric map until each sub-region contains one feature points. Then we can easily handle each sub-region and complete the alignment of all feature points.
In the third algorithm, we first present a theoretically robust, foldover-free 2D mesh warping algorithm. Then we apply this warping algorithm to texture mapping on 3D triangle meshes with hard positional constraints. The third algorithm can generate more pleasing visual representation, add fewer Steiner vertices on the 3D mesh embedding domain, and satisfy all user-specified constraints without foldovers.
|
|---|
