Shape Detection from Raw LiDAR Data with Subspace Modeling.

Area of Science:

• Geomatics and Geospatial Technology
• Computer Vision and Image Processing
• Urban Planning and Digital Twins

Background:

• LiDAR scanning is crucial for digitalizing large outdoor scenes, but raw data suffers from imperfections like missing regions, sampling density variations, and noise.
• These data imperfections pose significant challenges for advanced applications such as digital city modeling.
• Existing methods struggle to robustly handle the complexities of real-world LiDAR data.

Purpose of the Study:

• To develop a robust method for processing raw LiDAR data by addressing common imperfections.
• To enable accurate 3D urban scene modeling and shape detection from imperfect LiDAR scans.
• To enhance the utility of LiDAR data for higher-level digital city applications.

Main Methods:

• Proposed a novel approach based on locally classifying scan neighborhoods to model scene substructures.
• Introduced adaptive kernel-scale scoring, filtering, and clustering for substructure analysis.
• Developed a method to simultaneously recover local structures across all points, even with severe data imperfections.

Main Results:

• Demonstrated robust shape detection from raw LiDAR data by integrating local analyses.
• Successfully recovered local structures in the presence of significant noise, missing data, and sampling anisotropy.
• Validated the method's effectiveness and robustness on diverse and complex LiDAR datasets.

Conclusions:

• The proposed method offers a robust solution for processing imperfect LiDAR data, significantly improving 3D urban scene representation.
• This technique facilitates more ambitious applications in digital city modeling by overcoming data quality limitations.
• The approach shows strong potential for enhancing the accuracy and reliability of digital twins and geospatial analyses.