Feature-based multi-epoch rock slope monitoring using images and terrestrial laser scans

This study presents a comprehensive workflow for feature-based motion tracking of unstable rock slopes, exemplified at Mt. Hochvogel, using terrestrial imagery and terrestrial laser scanning (TLS). The workflow begins with multi-view images and TLS data acquired at different epochs. Features are detected both within each epoch, to reconstruct their 3D positions, and across epochs, to enable consistent motion tracking over time. TLS point clouds are integrated by projecting them into 2D image views, allowing a unified feature-based analysis across data sensors. The capability of this approach for monitoring over multiple acquisitions in the years 2018–2024 is evaluated using three different matching algorithms (SIFT, SuperPoint with LightGlue, and LoFTR). The results demonstrate that both image-based and TLS-based tracking can reliably capture slope movements. The evaluation against reference data captured with a total station shows that the results deviate on average by 1–6 mm in length and 4–7°in direction. Comparing the results from the two different sensors, the results show that image-based tracking generally achieves higher accuracy and more extensive coverage than tracking based on 2D representations derived from TLS point clouds. The analysis on Mt. Hochvogel shows that the main slope movement, characterized by a continuous opening of the large crevice at the summit, continued during the observation period. Apart from minor local movements, no significant acceleration or deceleration of the overall movement was observed. The spatial patterns indicate consistent block displacements of approximately 18–23 mm per year, with local variations in eroded or debris-covered areas. The image data used in this study are published as the HovoPhoto40 dataset (five epochs, 40 images each), providing a publicly available resource with potential for benchmarking photogrammetric monitoring workflows ( https://doi.org/10.5281/zenodo.16631844 ).