SEG-MAT: 3D Shape Segmentation Using Medial Axis Transform.

Area of Science:

• Computer Graphics
• Computational Geometry
• 3D Shape Analysis

Background:

• 3D object segmentation is crucial for computer graphics applications.
• Current methods struggle with complex geometry, high computation, and fragmented results due to limited global consideration.

Purpose of the Study:

• To develop an efficient and effective method for segmenting arbitrary 3D objects into structurally meaningful parts.
• To overcome limitations of existing 3D shape segmentation techniques.

Main Methods:

• The proposed method, SEG-MAT, leverages the medial axis transform (MAT) of 3D shapes.
• It utilizes the geometrical and structural information within the MAT to identify junctions between shape parts.

Main Results:

• SEG-MAT provides a simple and principled approach to 3D shape segmentation.
• The method effectively identifies junctions between different parts of a 3D shape.
• Extensive evaluations demonstrate superior segmentation quality compared to state-of-the-art methods.

Conclusions:

• SEG-MAT offers a significant improvement in 3D shape segmentation efficiency and quality.
• The method is approximately one order of magnitude faster than existing techniques.
• The medial axis transform provides rich information for effective 3D shape part segmentation.