Pairwise harmonics for shape analysis.

Area of Science:

• Computational Geometry
• Computer Graphics
• Geometric Modeling

Background:

• Traditional shape analysis relies on local surface point descriptors.
• Existing methods struggle to capture global geometric variations effectively.
• There is a need for more discriminative and efficient shape analysis techniques.

Purpose of the Study:

• To introduce a new shape analysis framework based on pairwise surface points.
• To develop novel shape descriptors that capture global geometric properties.
• To demonstrate the effectiveness of these descriptors in various geometry processing applications.

Main Methods:

• Developed a shape analysis framework utilizing descriptors based on pairs of surface points.
• Introduced new shape descriptors derived from isocurves of harmonic functions with extremal properties.
• Applied the pairwise descriptors to intrinsic reflectional symmetry axis computation, shape extremity matching, and surface segmentation/skeletonization.

Main Results:

• The pairwise shape descriptors are more global and discriminative than traditional methods.
• The proposed descriptors effectively capture underlying geometric variations.
• The framework leads to simpler and more efficient algorithms for tested applications.

Conclusions:

• Pairwise shape analysis offers a powerful alternative to point-based methods.
• The novel harmonic function-based descriptors provide robust geometric insights.
• This approach enhances the efficiency and simplicity of several key geometry processing tasks.