Deformable model retrieval based on topological and geometric signatures.

Area of Science:

• Computer Graphics
• Geometric Modeling
• Information Retrieval

Background:

• The increasing creation of 3D geometry models for applications like computer games necessitates efficient sharing and reuse techniques.
• Existing methods for 3D model matching and retrieval often struggle with deformable models (models in different poses) and are generally slow.

Purpose of the Study:

• To present a novel and efficient method for matching and retrieving 3D deformable models.
• To address the limitations of existing techniques in handling pose variations and speed.

Main Methods:

• Proposed Topological Point Ring (TPR) analysis to identify reliable topological points and rings within 3D models.
• Captured both local and global geometric information to characterize these topological features.
• Adapted Earth Mover Distance (EMD) for similarity comparison and constructed an indexing tree to accelerate retrieval.

Main Results:

• The developed method demonstrates improved accuracy and speed in matching and retrieving 3D deformable models.
• Experimental results validate the effectiveness of combining topological and geometric features.

Conclusions:

• The novel method offers an efficient solution for 3D deformable model retrieval, outperforming existing approaches in speed and accuracy.
• The integration of topological and geometric analysis provides a robust framework for handling complex 3D geometry data.