A synergistic integration of photogrammetry and Gaussian Splatting for back-in-time quality inspection

Photogrammetry is a well-established technique for 3D inspection of manufactured objects and machinery but its static nature limits re-inspection. If images lead to poor reconstruction, e.g., due to insufficient coverage, the object’s original state may become inaccessible, preventing further analysis. This work introduces a synergistic integration of photogrammetry and Gaussian Splatting (GS) to enable back-in-time quality inspection. GS creates “frozen” and fully navigable 3D digital scenes, i.e., high-fidelity 3D representations encoded with Gaussians, that preserve the spatial and temporal state of the object at the time of acquisition, allowing the generation of novel synthetic views (i.e., renders) to enrich the original photogrammetric reconstruction. A preliminary case study on the inlet lip skin part of the engine of an aircraft demonstrates how GS renders, when integrated into the photogrammetric workflow, improve the characterisation of surface defects. Results show enhanced defect localisation and a clearer delineation of damage geometry, suggesting that the proposed integration enables effective back-in-time inspections.