Two-Phase Oblique Photogrammetric Model for Automated Change Detection in Railroad Slopes

Abstract. Railroad slope detection is crucial for railroad inspection. However, traditional stability detection methods face challenges such as high costs, subjectivity, and reliance on prior information. To address these issues, we propose an automatic change detection algorithm based on a two-stage oblique photogrammetric model.The algorithm begins by extracting point cloud data from the structure and performs encryption and alignment preprocessing to eliminate spatial bias. It then dynamically selects core points to construct a cylindrical analysis domain, followed by comparing the differences in projected distances between the two phases of the point cloud against a preset threshold to identify changes.Experimental results demonstrate that our algorithm significantly outperforms traditional C2C and C2M methods in accurately detecting substantial changes, filtering out unrealistic alterations, adapting to various terrains, and reducing costs while enhancing efficiency. Notably, the algorithm achieves a maximum recognition accuracy of 96.825% at a threshold of 1 mm, underscoring its sophistication and effectiveness.