Approximate Intrinsic Voxel Structure for Point Cloud Simplification.

Area of Science:

• Computer Vision
• 3D Data Processing
• Geometric Modeling

Background:

• Point clouds are information-rich 3D representations but demand significant resources.
• High resource requirements limit point cloud applications in emerging fields.

Purpose of the Study:

• To propose a novel point cloud simplification method, Approximate Intrinsic Voxel Structure (AIVS).
• To address the diverse demands of real-world applications for efficient 3D data handling.

Main Methods:

• The AIVS method incorporates point cloud pre-processing, including denoising and down-sampling.
• It utilizes an AIVS-based approach for both isotropic and flexible simplification with intrinsic control of point distance.

Main Results:

• AIVS demonstrated significant advantages over existing methods in Moving Least Squares (MLS) surface approximation quality.
• The method excels in curvature-sensitive sampling, sharp-feature preservation, and processing speed.
• Experiments were conducted on Stanford, SHREC, and RGB-D scene models.

Conclusions:

• AIVS provides an effective solution for point cloud simplification, balancing accuracy and efficiency.
• The method's performance advantages make it suitable for various real-world 3D data applications.
• Source code is publicly available for further research and application.