Interactive CAD Reverse Modeling via Variational Curve Approximation.

Area of Science:

• Computer-Aided Design (CAD)
• Reverse Engineering
• Geometric Modeling

Background:

• Reconstructing editable CAD models from mesh data is a significant challenge in reverse engineering.
• Current methods often compromise geometric accuracy or topological consistency, especially for complex meshes with analytic design intent.

Purpose of the Study:

• To present an interactive CAD reverse modeling framework that unifies mesh segmentation and parametric reconstruction.
• To improve the accuracy, consistency, and automation of CAD model reconstruction from mesh data.

Main Methods:

• Utilizes variational curve approximation to extract and refine surface-aligned cutting curves from segmented mesh regions.
• Incorporates primitive attributes and geometric constraints into the refinement process.
• Employs coplanar profile detection and voxel-similarity-based Boolean inference for topological order restoration.

Main Results:

• Achieves automatic recovery of closed loops, constraint relations, and operation parameters, resulting in compact, topologically consistent CAD representations.
• Demonstrates superior geometric fidelity and structural consistency compared to existing state-of-the-art techniques.
• Produces watertight and topologically consistent models suitable for downstream CAD design.

Conclusions:

• The proposed framework offers a robust solution for CAD reconstruction from mesh data, addressing limitations of current approaches.
• The method enhances automation and fidelity, making reconstructed models readily applicable for practical design tasks.
• This work advances the field of reverse engineering by providing a unified and effective approach to mesh-to-CAD conversion.