A Point Cloud Simplification Algorithm Based on Weighted Feature Indexes for 3D Scanning Sensors.

Area of Science:

• Computer Science
• Geographic Information Systems
• Computational Geometry

Background:

• Conventional point cloud simplification methods suffer from issues like non-uniformity, poor characteristic reflection, and high computational complexity.
• Existing algorithms often struggle with effective feature point identification and weight distribution, leading to suboptimal simplification results.

Purpose of the Study:

• To introduce a novel Multi-Index Weighting Simplification Algorithm (MIWSA) for enhanced point cloud simplification.
• To address the limitations of traditional algorithms by improving accuracy, uniformity, and computational efficiency.

Main Methods:

• Organizing point clouds using bounding boxes and kd-trees to identify point neighborhoods and segment data.
• Calculating feature indices for each point to quantify characteristics.
• Employing the Analytic Hierarchy Process (AHP) and Criteria Importance Through Intercriteria Correlation (CRITIC) for index weighting and feature point determination.
• Classifying non-feature points for saving or abandonment based on spatial relationships with feature points.

Main Results:

• The MIWSA demonstrated a 5% to 15% improvement in overall accuracy compared to existing algorithms on 3D model and field area scanning datasets.
• The algorithm achieved shorter running times than most conventional methods.
• Evaluations using surface area, patch numbers, and DEM error statistics confirmed the algorithm's effectiveness.

Conclusions:

• The MIWSA effectively simplifies point clouds with greater precision and uniformity.
• This algorithm offers a significant advancement in point cloud processing, balancing accuracy and efficiency.
• MIWSA provides a robust solution for applications requiring high-fidelity point cloud simplification.