Feasibility of Action Camera-Based Videogrammetry for Multi-Temporal 3D Monitoring of Rubble-Mound Breakwaters

Coastal protection infrastructures such as rubble-mound breakwaters (RMBs) demand frequent geometric inspection to quantify armor-layer dynamics and support reproducible structural monitoring. While UAV-based photogrammetry and LiDAR are established reference techniques for rapid 3D mapping, high revisit rates remain operationally constrained by wind sensitivity, sensor payload limits, and regulatory flight restrictions. Videogrammetry complements these approaches by increasing inter-frame overlap and mitigating missed-trigger acquisitions, especially useful in complex coastal scenes (e.g., those affected by occlusions between armor units and block interstices). As in conventional photogrammetry, videogrammetry relies on image redundancy and self-calibration rather than highly sophisticated instrumentation. Despite this potential, consumer-grade action cameras remain scarcely validated for multi-epoch 3D monitoring in coastal engineering, mainly due to wide-angle lens distortion and coarse onboard GNSS geotag precision.

This study assesses pole-mounted GoPro videogrammetry for multi-temporal 3D relative change detection in the emerged portions of a detached rubble-mound breakwater at Cabedelo do Douro (PT). Two survey epochs were acquired in July 2024 and November 2024 to characterize the above-water zone, inspecting the seaward slope, the landward armor-toe transition, and the horizontal crest platform segment at one of the heads of the RMB. Frames were extracted at 1 Hz and processed in Metashape using an SfM-MVS (Structure-from-Motion Multi-View Stereo) self-calibrating camera model. Multi-epoch point clouds were coregistered in CloudCompare with ICP (Iterative Closest Point) refinement over stable crest and toe areas, and 3D changes were quantified using M3C2 (Multiscale Model-to-Model Cloud Comparison), generating signed distance maps and detection histograms. A concurrent UAV-RTK survey, supported by additional GNSS-measured ground control points (GCPs), served as a geometric benchmark.

Mean ActionCam-to-UAV sensor offsets were +0.06 m, confirming that, despite potentially unstable absolute georeferencing in GoPro-derived reconstructions, the resulting point clouds preserve sufficient geometric and scale consistency to support relative multi-temporal 3D change detection and the identification of concrete armor-unit displacements. Results confirm that pole-mounted videogrammetry supports rapid, repeatable, low-cost SHM (Structural Health Monitoring) observations, providing defensible detection thresholds and reproducible change-detection limits for engineering interpretation and maintenance support.