Automated construction of low-resolution, texture-mapped, class-optimal meshes

Area of Science:

• Computer Graphics
• Computational Geometry
• 3D Shape Analysis

Background:

• Groupwise processing of 3D meshes is crucial for tasks like modeling shape variation and tracking surface motion.
• Existing mesh processing tools often operate on individual meshes, lacking efficiency for large datasets with dense correspondence.

Purpose of the Study:

• To present a novel framework for groupwise processing of 3D meshes in dense correspondence.
• To extend existing mesh processing techniques to operate efficiently on sets of meshes.
• To enable the construction of low-resolution 3D morphable models from groupwise processed data.

Main Methods:

• Developed a geodesic-based surface flattening and spectral clustering algorithm for class-optimal flattening.
• Modified an iterative edge collapse algorithm for groupwise mesh simplification while preserving correspondence.
• Introduced methods for computing class-optimal texture coordinates for simplified meshes.
• Presented alternative algorithms for topologically symmetric data, yielding symmetric flattening and low-resolution mesh topology.

Main Results:

• Demonstrated successful flattening, simplification, and texture mapping on three diverse datasets.
• Showcased the framework's ability to construct low-resolution 3D morphable models.
• Validated the effectiveness of groupwise processing for handling sets of meshes in dense correspondence.

Conclusions:

• The proposed groupwise mesh processing framework effectively handles sets of meshes in dense correspondence.
• The developed algorithms facilitate the creation of efficient and accurate low-resolution 3D morphable models.
• This approach advances 3D shape analysis and surface motion tracking by enabling groupwise operations.