Smartphone-based photogrammetry for synthetic surface crack detection in reinforced concrete bridge structures

Surface cracking in reinforced concrete bridges is a critical indicator of structural deterioration. Recent studies merge computer vision and point cloud data for crack detection but require scientific-grade instrumentation and aerial data. This study investigates a citizen science approach: smartphone-based photogrammetry as a low-cost solution for capturing and interpreting such damage under real lighting conditions. Smartphone photogrammetric models were created and analysed through visual inspection, RGB colour variance, point cloud alignment via Iterative Closest Point (ICP), and lightweight image-based detection algorithms, including thresholding, subtraction, structural similarity index measure (SSIM), and edge detection. Results show that, with optimised image capture and prioritising lighting, angle diversity, and overlap, synthetic surface cracks simulating shear and flexural failures are detectable. While RGB variance shows diagnostic limitations due to environmental sensitivity, ICP-enabled spatial coherence proved crucial for comparative analysis. While direct thresholding yielded more interpretable features in a qualitative setting, it is observed that quantitative algorithmic performance remains mixed depending on the evaluation metric. Notable examples include direct thresholding with better precision findings, whereas SSIM and edge detection indicate better recall rates. With a structured workflow, consumer-grade devices offer a scalable alternative to traditional reconnaissance methods.