Highly parallel algorithms for visual-perception-guided surface remeshing.

Area of Science:

• Computer Graphics
• Geometric Modeling
• Computational Geometry

Background:

• Polygonal models often have complex connectivity, posing challenges for traditional remeshing algorithms.
• Existing methods may struggle with robustness and efficient parallelization for large datasets.

Purpose of the Study:

• To develop a robust and efficient framework for remeshing polygonal models.
• To leverage GPU parallelization for accelerated geometric processing.
• To enable high-fidelity mesh generation and model simplification.

Main Methods:

• Employs mesh-free techniques for processing surface sample points.
• Extracts visual-perception information in image space and maps it to Euclidean space.
• Generates a saliency field for resampling and uses a projection operator for optimization.
• Utilizes GPU parallelization for all algorithmic components.

Main Results:

• Successfully remeshed diverse polygonal models into well-shaped triangular meshes.
• Achieved high visual fidelity in the generated meshes.
• Demonstrated robustness to input models with problematic connectivity.
• Framework effectively performs model simplification by controlling vertex count.

Conclusions:

• The proposed framework offers an efficient and robust solution for polygonal model remeshing.
• GPU parallelization significantly enhances processing speed.
• The approach is suitable for both high-fidelity mesh generation and model simplification tasks.