Ofir Engolz Rony Goldenthal Dani Lischinski Daniel Cohen-Or
Stages of progressive encoding of David model (24085 vertices).
Top row (from left): 5 bit quantization of coarse mesh levels (154 , 494 , 1786 and 6402 vertices).
Bottom row: Fine mesh with 5,7 and 9 bit quantization (24085 vertices). The last image is the original model
Proceedings of the 5th Korea Israel conference on Geometric Modeling and Computer Graphics, 2004
Abstract
Today’s huge irregular polygon meshes require effective compression techniques to reduce the associated storage requirements, network bandwidth and transmission times. In this paper, we describe a new method for compressing the geometry of an irregular triangle mesh, which is both scalable (encoding and decoding time is linear in the number of mesh vertices) and progressive, enabling the decoder to generate a progression of approximations while the data is being transmitted. The encoding approach utilized by our method is based on the recently introduced High-Pass Quantization technique. The decoding stage, however, uses a linear AlgebraicMulti- Grid solver, which makes high-pass quantization truly scalable. The speed of our solver makes it possible to update the solution as more bits are received, resulting in a progressive transmission scheme.
Paper
IK2004 Presentation
BibTeX Entry:
@InProceedings{AMG_HPQ04, author = {Ofir Engolz and Rony Goldenthal and Dani Lischinski and Daniel Cohen-Or}, title = {An Algebraic Multi-Grid Approach for High-Pass Quantization}, booktitle = {Proceedings of The 5th Korea-Israel Bi-National Conference on Geometric Modeling and Computer Graphics}, pages = {57--62}, location = {Seoul, Korea}, year = { 2004} }Happy Budha High pass quantization and naive quantization
Left: the Original Happy Budha model, middle our 9bit High Pass quantization and on the right a naive 9 bit quantization
Progressive transmission of Venus and Horse models
