Spectral pattern classification in lidar data for rock identification in outcrops.

Area of Science:

• Geology
• Computer Science
• Remote Sensing

Background:

• Terrestrial laser scanning (TLS) generates dense point clouds for geological mapping.
• Classifying spectral signatures in TLS data is crucial for rock identification and digital outcrop modeling.
• Existing methods may lack automated spectral signature analysis for diverse rock types.

Purpose of the Study:

• To develop and implement a novel method for detecting and classifying spectral signatures in TLS point clouds.
• To create a software tool (K-Clouds) for automated rock identification and digital outcrop model generation.
• To enable geologists in better understanding rock composition and identifying weathering-induced changes.

Main Methods:

• Development of K-Clouds software utilizing cluster analysis on point cloud intensity return values.
• Histogram analysis of point cloud data to guide the classification process.
• User-defined class indication for processing spectral intensity data.

Main Results:

• Successful detection and classification of spectral signatures corresponding to different rock types in outcrops.
• Generation of classified point clouds facilitating geological interpretation.
• Identification of subtle physical-chemical rock changes attributed to weathering and compositional variations.

Conclusions:

• The K-Clouds software provides an effective tool for automated rock classification from TLS data.
• This method enhances geological interpretation of outcrops and aids in understanding rock alteration processes.
• The approach supports the creation of detailed digital outcrop models with improved geological insights.