Surface reconstruction based on CAD model driven priori templates.

Area of Science:

• Computer Graphics
• Geometric Modeling
• Reverse Engineering

Background:

• Traditional 3D reconstruction methods directly process scanned point clouds, often leading to issues with feature preservation and detail recovery.
• Existing techniques struggle with high-quality surface reconstruction, particularly for complex shapes with sharp features.

Purpose of the Study:

• To propose a novel method for reconstructing high-quality 3D models from scanned point clouds using prior templates.
• To improve upon existing reconstruction techniques by incorporating Computer-Aided Design (CAD) models for guidance.
• To achieve accurate feature preservation and detail recovery in the reconstructed 3D models.

Main Methods:

• A priori template is constructed from a CAD model, defining basic geometric elements as units for reconstruction.
• Point cloud data corresponding to each basic element is extracted from the original scan.
• Gaussian weighted averaging is used to determine actual geometric parameters by searching nearby points.
• Local reconstruction of geometric elements is performed using Non-Uniform Rational B-Splines (NURBS) surface fitting.
• The complete 3D model is obtained by integrating these locally reconstructed elements.

Main Results:

• The proposed method successfully reconstructs 3D models with high fidelity.
• Sharp features and intricate details from the original scanned data are effectively preserved.
• The approach demonstrates superior performance compared to direct point cloud fitting methods.

Conclusions:

• The template-guided 3D reconstruction method offers a robust solution for accurate surface reconstruction.
• This technique effectively addresses the challenges of quality, feature preservation, and detail recovery in 3D modeling.
• The integration of CAD models as prior templates significantly enhances the reconstruction process.