Blending Surface Segmentation and Editing for 3D Models.

Area of Science:

• Computer Graphics
• Computer Vision
• Geometric Modeling

Background:

• Existing methods for 3D model recovery struggle with identifying blending surfaces.
• Blending surfaces are crucial transitional regions between larger surface patches in 3D models.

Purpose of the Study:

• To develop a novel approach for automatic segmentation and surface fitting of 3D models, particularly mechanical parts.
• To accurately recover geometric parameters, including those for blending surfaces.

Main Methods:

• Formulating structural segmentation as a Markov Random Field (MRF) labeling problem.
• Developing a new clustering algorithm to create superfacets using local 3D geometric information, identifying quadric and rolling-ball blending regions.
• Applying a specialized MRF framework for partitioning models into meaningful patches.
• Using an iterative optimization algorithm based on skeleton extraction to fit rolling-ball blending patches.

Main Results:

• The proposed method effectively segments 3D models and accurately fits surfaces, including challenging blending regions.
• Experiments demonstrate the robustness and effectiveness of the approach on complex models.
• The method shows superiority compared to state-of-the-art techniques.

Conclusions:

• The novel MRF-based approach successfully addresses the limitation of identifying blending surfaces in 3D models.
• The algorithm offers accurate geometric parameter recovery for various surface types.
• The method has practical applications in areas like mesh editing.