A novel multi-camera DIC without point cloud stitching using global co-visible relationships

Multi-camera digital image correlation (MC-DIC) is an effective and practical image-based optical technique for large field-of-view or multi-surface deformation field measurements. Conventional MC-DIC methods typically require individual analysis of images captured by each stereo-DIC system before aligning the results into a global coordinate system. Due to variations in calibration accuracy or missing parameters across individual systems in practical measurements, discrete measurement results may exhibit noticeable gaps when stitched together or cannot be unified into the same coordinate system. To address these problems, this work proposes a novel MC-DIC method based on global co-visible relationships. With each stereo-rig calibrated, the classic SURF feature matching algorithm extracts and matches feature points between views, determining local co-visible relationships. By using the Union-Find algorithm to organize the local co-visible relationships in the multi-camera system, mappings of 3D points to each 2D reference image are established, i.e., global co-visible relationships. Then, the coordinate systems of multiple stereo-rigs can be aligned based on these relationships and the PnP algorithm. In the unified coordinate system, obtaining 3D points through multi-view stereo vision reconstruction allows for subsequent optimization and analysis without individual computations and point cloud stitching. The feasibility and advantages of the proposed method are demonstrated through two practical experiments.