Surface Registration with Eigenvalues and Eigenvectors.

Area of Science:

• Computational geometry
• Computer vision
• Medical imaging analysis

Background:

• Accurate surface registration is crucial for analyzing complex shape variations.
• Existing methods struggle with non-isometric deformations and computational efficiency.

Purpose of the Study:

• To develop a novel spectral surface registration technique for accurate localization of non-isometric deformations.
• To improve computational efficiency while maintaining registration accuracy.

Main Methods:

• Mapping eigenvalues and eigenvectors of the Laplace-Beltrami operator of shapes.
• Optimizing an energy function integrating eigenvalue smoothness and eigenvector alignment at feature points.
• Generating feature points using static points of specific eigenvectors for computational efficiency.

Main Results:

• The proposed method achieves accurate surface registration and localization of non-isometric deformations.
• Utilizing eigenvalues and eigenvectors at feature points significantly improves computational efficiency without sacrificing accuracy.
• The method accurately defines vertex displacement, outperforming non-rigid Iterative Closest Point (ICP) and other spectrum-based methods.

Conclusions:

• The spectral surface registration technique offers a robust and efficient solution for analyzing complex shape variations.
• The method demonstrates significant advantages in accuracy and computational performance on synthetic and real-world datasets.
• This approach facilitates precise analysis of surface deformations in applications like medical imaging.