Multi-camera stereo-digital image correlation: a review

The core idea of multi-camera stereo-DIC is to extend conventional binocular single-FOV measurement into a distributed full-field measurement framework, in which multiple subsystems collaboratively acquire data and represent the results in a unified global coordinate system. As described in the review, the basic workflow consists of dividing the target surface into multiple subregions, capturing them separately with multiple synchronized stereo-DIC subsystems, and then transforming the results from the local coordinate system of each subsystem into a unified global coordinate system. This strategy of “measuring separately and integrating globally” enables multi-camera stereo-DIC to expand the measurement area while maintaining high measurement accuracy, thereby alleviating the classical trade-off between measurement coverage and measurement uncertainty in conventional stereo-DIC.

From the perspective of technical implementation, the review further classifies multi-camera stereo-DIC into two categories. The first integrates the local results obtained by multiple binocular stereo-DIC subsystems into a unified global coordinate system, whereas the second relies on multiview geometry for direct three-dimensional reconstruction. The former is more adaptable to complex curved surfaces and full-surface measurements, while the latter offers advantages in avoiding ambiguity caused by multi-valued results in overlapping regions and in improving 3D reconstruction accuracy. Based on these two technical routes, the review summarizes key advances in measurement-area expansion, measurement-uncertainty control, multi-scale data integration, high-dynamic-range imaging, and high-speed measurement, showing that multi-camera stereo-DIC is evolving from a method primarily designed for large-FOV measurement into a multi-capability platform for complex measurement tasks.

Beyond methodological advances, the review also shows that multi-camera stereo-DIC has been applied in a variety of representative experimental scenarios. For large-FOV measurements of large structures, this technique has been used to measure the three-dimensional dynamic deformation of a box-type structure on a large shaking table (Fig. 3). Using two coordinated stereo-DIC subsystems together with a real-time extrinsic-parameter correction method, the system achieved submillimeter-level displacement measurement accuracy and reliable acceleration analysis. This demonstrates that multi-camera systems can not only “see more broadly,” but also maintain stable and reliable measurement performance under strong vibration conditions.

Moreover, conventional displacement sensors are often unable to capture the complex spatial displacements involved in progressive collapse, especially when critical nodes are widely distributed and on-site measurement is difficult. To address this issue, researchers incorporated UAV-assisted close-range photogrammetry into a multi-camera system, enabling displacement monitoring during both the tensioning construction and progressive-collapse processes of a suspen-dome structure. Experimental validation showed that the system achieved a displacement measurement accuracy of 0.05 mm, with a relative error of about 0.02%. In the progressive-collapse test, the system successfully captured instantaneous nodal displacements and provided reliable data for failure-mode analysis and numerical simulation (Fig. 4).

Meanwhile, the review also highlights the unique advantages of multi-camera stereo-DIC in panoramic measurements. For cylindrical or circumferentially closed structures, multiple cameras can be arranged around the specimen, with adjacent cameras forming stereo measurement subsystems and overlapping fields of view enabling continuous circumferential measurement. Related studies have shown that such systems have been used for 360° full-field displacement measurement and, in mechanical experiments, have supported tasks such as through-thickness strain measurement, true stress–strain curve estimation, and anisotropic parameter identification. This indicates that multi-camera stereo-DIC is evolving from local FOV expansion toward full-surface measurement, with its application boundaries continuing to broaden.

The review concludes that multi-camera stereo-DIC is driving digital image correlation from a conventional local measurement technique toward a distributed three-dimensional full-field measurement platform for complex objects and challenging environments. It not only alleviates the trade-off between measurement area and measurement uncertainty in conventional stereo-DIC, but also continues to extend its capabilities in large-FOV, panoramic, multi-scale, high-dynamic-range, and high-speed measurements. Looking ahead, further advances are still needed in heterogeneous camera collaboration, real-time data transmission and computation, robustness in complex environments, and deep-learning-enabled end-to-end measurement. With the continued development of intelligent sensing, parallel computing, and artificial intelligence methods, multi-camera stereo-DIC is expected to play an increasingly important role in large-scale structural monitoring, extreme-environment experiments, and advanced manufacturing inspection.