Fast and effective feature-preserving mesh denoising.

Area of Science:

• Computer Graphics
• Computational Geometry
• Image Processing

Background:

• Mesh denoising is crucial for processing noisy 3D data.
• Existing methods often struggle to preserve sharp features or require manual parameter tuning.
• There is a need for efficient and robust mesh denoising algorithms.

Purpose of the Study:

• To develop a simple and fast mesh denoising method.
• To effectively remove noise while preserving critical mesh features like sharp edges and corners.
• To improve upon the speed and ease of implementation of existing surface denoising techniques.

Main Methods:

• A two-stage approach involving iterative filtering of noisy face normals using a simplified trimmed quadratic weight function.
• Iterative updating of vertex positions based on denoised face normals and a least-squares error criterion.
• Automatic determination of iteration step size using gradient descent, eliminating the need for user input.

Main Results:

• The proposed method effectively removes noise from 3D meshes.
• Preservation of mesh features, including sharp edges and corners, is achieved.
• The algorithm demonstrates significant speed and simplicity advantages over previous methods.
• Convergence of the vertex position updating approach is mathematically proven.

Conclusions:

• The presented mesh denoising method offers an effective, fast, and simple solution for noise reduction.
• Its ability to preserve mesh features makes it suitable for various computer graphics applications.
• The automatic iteration step size determination and proven convergence enhance its practical utility and reliability.