Volume Change Analysis in Environmental Point Clouds

Abstract

Observation of landscape change is essential for strategic planning, risk assessment, and disaster prevention in volatile environments, especially on steep rocky slopes and cliffs where rockfalls can endanger human lives. To support the analysis of large point cloud scans of these environments, we propose a novel method for estimating local terrain changes and computing corresponding material displacement by combining existing change detection algorithms, such as M3C2, with the identification of smaller areas of significant change and the construction of localized watertight meshes. Our approach processes complex terrain to extract and visualize critical information, providing more detailed volumetric analysis than standard software, which is often limited to point coloring. We evaluate our method on various simple synthetic scenarios, including hill, cube, and sphere tests, as well as a real UAV dataset of coastal cliffs captured over multiple years. The experimental results show that while the method can become unstable with complex geometry or noisy data, it produces plausible volume displacement values and demonstrates high potential for more efficient terrain deformation analysis. We conclude that our proposed pipeline provides a strong foundation for advanced geomorphological observation and the development of automated systems for land risk assessment.