From Point Clouds to Cultural Landscapes: Open-Source Machine Learning Applications for Archaeological UAV LiDAR segmentation

Nicodemo Abate, Alessia Frisetti, Gabriele Ciccone, Antonio Minervino Amodio, Maria Sileo, Rosa Lasaponara, Nicola Masini

Abstract

This study presents an open-source methodological workflow for processing Unmanned Aerial System (UAS) LiDAR data using a probabilistic machine learning algorithm to enhance the visibility and detection of archaeological features under vegetation. The proposed framework combines the 3DMASC plugin for CloudCompare with the Relief Visualization Toolbox (RVT) and QGIS to deliver an accessible, non-programmer-friendly solution for point cloud classification and derivative model enhancement. The methodology is validated through two case studies: the Kastrì-Pandosia site in Epirus, Greece, and Torre Castiglione in Apulia, Italy. Both sites, obscured by dense vegetation, revealed critical archaeological structures—including defensive walls, terraces, and ancient routes—following segmentation and visualization. Results confirm the robustness and replicability of the approach, reinforcing the value of open-source strategies in archaeological remote sensing.

Download

IMEKO-Metroarchaeo-2025-042.pdf

DOI

10.21014/tc26-2025.042

IMEKO TC

TC26 - Metrology for Cultural Heritage

Back to the proceedings