Zerotree Coding of Subdivision Wavelet Coefficients in Dynamic Time-Varying Meshes

Area of Science:

• Computer Vision
• Computer Graphics
• Signal Processing

Background:

• Dynamic mesh coding is crucial for representing 3D geometry over time.
• Existing Video-based Dynamic Mesh Coding (V-DMC) methods lack quality scalability.
• Efficient compression of time-varying 3D mesh geometry is an ongoing challenge.

Purpose of the Study:

• To develop a progressive coding system for mesh geometry with quality scalability.
• To integrate a novel zerotree coding approach into the V-DMC framework.
• To enable efficient and versatile compression of dynamic 3D meshes.

Main Methods:

• Proposed an alternative method for encoding subdivision wavelet coefficients.
• Utilized a zerotree coding approach operating directly in the 3D mesh space.
• Identified parent-child relationships across wavelet coefficient subdivision levels.

Main Results:

• Achieved elegant, visually attractive lossy-to-lossless mesh geometry reconstruction.
• Demonstrated comparable bitrate to V-DMC displacement coding for lossless compression.
• Enabled multiple embedded quality resolution levels within a single bitstream.

Conclusions:

• This is the first zerotree-based method for dynamic time-varying mesh compression.
• This work presents the first embedded quality-scalable approach for the V-DMC framework.
• The proposed system offers enhanced versatility and quality scalability for dynamic mesh coding.