M. Ohuchi and T. Saito (Japan)
Geometric Modeling, Surface Representation, Subdivision Surface, Directional Displacement, Data Compression
Advances in range image acquisition and integration allow us to construct geometric models of complex objects from highly detailed and densely sampled data obtained from geometry scanning. The problem of geometric modeling is how to construct accurate and compact models. In this paper, we present a geometric modeling based on the shape representation which consists of coarse geometry and fine surface details. The proposed modeling is accomplished by converting dense irregular polygon meshes of arbitrary topology into semi-regular triangular meshes. For representing the coarse geometry, the original mesh is simplified using a sequence of vertex removal prioritized to preserve the global structure into coarse mesh called base mesh, and Loop subdivision scheme is successively applied to the base mesh to create a domain mesh as a smooth approximation of original mesh. For representing surface details, we propose directional displacements over domain mesh. Based on the prescribed distance measure, the directional displacements are determined so as to follow the surface details of the original meshes as faithfully as possible. To make the representation simple and compact, we consider data compression schemes for directional displacements using run-length and Huffman coding. Several examples using well known data sets are shown to demonstrate the accuracy and compactness of the proposed modeling compared to the conventional modeling ”displaced subdivision surface”.
Important Links:
Go Back
