Framework for mersuring deformation of cylindrical sample in 2D DIC

Digital image correlation (DIC) is a powerful tool in lab-scale rock mechanics, yet its application to cylindrical samples is compromised by perspective-induced distortions. These optical effects lead to significant inaccuracies in measuring deformations, especially impacting the reliability of Poisson's ratio calculations. To address this, we developed a specialised preprocessing workflow to rectify raw images before correlation.

The proposed method uses a custom Python script that performs image denoising, camera calibration, and lens distortion correction and an unwrapping algorithm that projects the cylindrical surface onto a 2D plane, effectively "flattening" the sample geometry. This allows standard 2D DIC software, such as NCORR, to process the data without the geometric bias inherent in radial perspectives.

To validate the workflow, results were benchmarked against a 3D-DIC system and physical sensors. Preliminary data shows that our rectification process significantly improves displacement accuracy on lateral surfaces, providing a low-cost yet precise alternative to complex 3D setups. This enhancement is crucial for characterising displacement over the full sample surface where traditional strain gauges are limited. Future work will focus on refining pixel-level interpolation to further minimise noise in high-strain zones.