Spatially-adaptive bases in wavelet-based coding of semi-regular meshes

Leon Denis, Ruxandra-Marina Florea, Adrian Munteanu, Peter Schelkens

Onderzoeksoutput: Conference paper

Samenvatting

In this paper we present a wavelet-based coding approach for semi-regular meshes, which spatially adapts the employed wavelet basis in the wavelet transformation of the mesh. The spatially-adaptive nature of the transform requires additional information to be stored in the bit-stream in order to allow the reconstruction of the transformed mesh at the decoder side. In order to limit this overhead, the mesh is first segmented into regions of approximately equal size. For each spatial region, a predictor is selected in a rate-distortion optimal manner by using a Lagrangian rate-distortion optimization technique. When compared against the classical wavelet transform employing the butterfly subdivision filter, experiments reveal that the proposed spatially-adaptive wavelet transform significantly decreases the energy of the wavelet coefficients for all subbands. Preliminary results show also that employing the proposed transform for the lowest-resolution subband systematically yields improved compression performance at low-to-medium bit-rates. For the Venus and Rabbit test models the compression improvements add up to 1.47 dB and 0.95 dB, respectively.
Originele taal-2English
TitelSPIE Photonics Europe. Optics, Photonics, and Digital Technologies for Multimedia Applications
Aantal pagina's8
Volume7723
StatusPublished - 12 apr 2010
EvenementUnknown - Stockholm, Sweden
Duur: 21 sep 200925 sep 2009

Conference

ConferenceUnknown
Land/RegioSweden
StadStockholm
Periode21/09/0925/09/09

Vingerafdruk

Duik in de onderzoeksthema's van 'Spatially-adaptive bases in wavelet-based coding of semi-regular meshes'. Samen vormen ze een unieke vingerafdruk.

Citeer dit