Scalable L-infinite coding of meshes.

Area of Science:

• Computer Graphics
• Digital Signal Processing
• Geometric Modeling

Background:

• Traditional mesh encoding often uses mean-square error (L-2 distortion).
• Existing methods lack precise control over local geometric errors in decoded meshes.
• Scalability and animation features are critical in 3D object encoding systems.

Purpose of the Study:

• To propose and evaluate a novel L-infinite mesh-coding approach for local-error control.
• To ensure a guaranteed upper bound on the maximum vertex position error.
• To achieve L-infinite scalability in 3D mesh geometry encoding.

Main Methods:

• Development of a wavelet-based L-infinite-constrained coding algorithm for meshes.
• Implementation and testing within the MESHGRID scalable 3D object encoding system (MPEG-4 AFX).
• Experimental comparison of L-infinite coding against L-2-oriented approaches.

Main Results:

• The proposed system guarantees a predictable upper bound on L-infinite distortion.
• Demonstrated advantages of L-infinite coding over L-2 coding in terms of local error control.
• Achieved fast, real-time rate allocation implementation.

Conclusions:

• The L-infinite mesh-coding approach effectively guarantees an upper bound on local mesh errors.
• The method preserves scalability and animation capabilities of existing codecs.
• This approach offers a superior alternative for applications requiring strict local error bounds.