Summary: | 博士 === 國立臺灣科技大學 === 電子工程系 === 96 === In this dissertation, we propose two mesh processing techniques for 3-D graphics-related applications. First, to automatically segment a 3-D mesh-based object, we propose a mesh decomposition scheme based on PCA (Principal Component Analysis) and Boolean operations. It is well known that most of the existing 3-D mesh-based decomposition schemes are bottom-up approaches. With a bottom-up decomposition strategy, it is difficult to devise a systematic way to determine the main body of an arbitrary 3-D object. In this work, we combine the 3-D coordinates and the protrusion degrees of the dual vertices of a 3-D mesh object to form a set of 4-D feature vectors. Then, we perform PCA on the set of 4-D feature vectors derived from the 3-D mesh-based object. After PCA transformation, we can identify all the salient components of an arbitrary 3-D object and precisely separate the salient components of the 3-D object from its main body automatically.
Second, to improve the search/retrieval process of 3-D object in the network environment, we propose a compact 3-D object representation scheme which transforms a 3-D object from the original space into a new coordinate frame by using the Isomap (Isometric feature mapping) method. The transformation process preserves the structure of an object’s salient parts as well as the geometrical relationships between the parts. From the viewpoint of cognitive psychology, the data distributed on the Isomap manifold can be regarded as a set of significant features of a 3-D mesh-based object. To perform efficient matching, we project the Isomap domain’s 3-D object onto two 2-D maps. We then use the two 2-D feature descriptors as the basis for measuring the degree of similarity between two arbitrary 3-D mesh-based objects. The results of experiments demonstrate that the proposed method is very effective in retrieving similar 3-D models.
|