Tensor Maps for Synchronizing Heterogeneous Shape Collections.

Area of Science:

• Computer Graphics
• Computational Geometry
• Geometric Modeling

Background:

• Establishing high-quality correspondence maps is crucial for managing geometric shape collections.
• Existing map synchronization methods require homogeneous shape collections and correct pairwise maps.

Purpose of the Study:

• To develop a robust multiple map synchronization approach for heterogeneous shape collections.
• To output consistent dense pairwise shape maps simultaneously.

Main Methods:

• Utilized a novel tensor-based representation for map synchronization.
• Developed an approach that handles heterogeneous shape collections.
• Simultaneously synchronizes multiple maps.

Main Results:

• The tensor-based representation is more efficient and robust than previous matrix-based methods.
• Demonstrated effectiveness across diverse geometric shape datasets.
• Successfully applied to shape clustering and co-segmentation tasks.

Conclusions:

• The proposed tensor-based multiple map synchronization method effectively addresses limitations of prior work.
• Enables consistent and accurate map generation for heterogeneous shape collections.
• Offers practical applications in shape analysis and organization.