Compression of mesh attributes becomes a challenging problem due to the greatneed for efficient storage and fast transmission. This paper presents a novel geometric signalcompression framework for all mesh attributes,...Compression of mesh attributes becomes a challenging problem due to the greatneed for efficient storage and fast transmission. This paper presents a novel geometric signalcompression framework for all mesh attributes, including position coordinates, normal, color,texture, etc. Within this framework, mesh attributes are regarded as geometric signals defined onmesh surfaces. A planar parameterization algorithm is first proposed to map 3D meshes to 2Dparametric meshes. Geometric signals are then transformed into 2D signals, which are sampled into 2Dregular signals using an adaptive sampling method. The JPEG2000 standard for still imagecompression is employed to effectively encode these regular signals into compact bit-streams withhigh rate/distortion ratios. Experimental results demonstrate the great application potentials ofthis framework.展开更多
文摘Compression of mesh attributes becomes a challenging problem due to the greatneed for efficient storage and fast transmission. This paper presents a novel geometric signalcompression framework for all mesh attributes, including position coordinates, normal, color,texture, etc. Within this framework, mesh attributes are regarded as geometric signals defined onmesh surfaces. A planar parameterization algorithm is first proposed to map 3D meshes to 2Dparametric meshes. Geometric signals are then transformed into 2D signals, which are sampled into 2Dregular signals using an adaptive sampling method. The JPEG2000 standard for still imagecompression is employed to effectively encode these regular signals into compact bit-streams withhigh rate/distortion ratios. Experimental results demonstrate the great application potentials ofthis framework.
