Evaluation of Processing Parameters in High-Speed Digital Image Correlation for Strain Measurement in Rock Testing.

High-speed digital image correlation (DIC) offers a powerful tool for full-field measurement of transient deformation of rock. One of the critical components controlling the measurement accuracy in DIC is the processing parameter, but neither manufacturers of DIC supplier nor the literature provides guidelines for optimal selection of processing parameter in high-speed DIC for rock testing. In this work, effect of six processing parameters on the accuracy of DIC-measured strain is evaluated through a dynamic compression test on sandstone. The processing parameters, including area of interest (AOI) size, correlation criterion, subset size, step size, filtering and incremental correlation are evaluated by comparing the error between DIC-measured and theoretical overall strain history. The quality of strain field obtained with different processing parameters is also assessed. Results show that the overall strain averaged from the virtual strain gauge (VSG) of AOI covering more than 43% of the specimen length is reliable. To characterize reasonable strain field, the VSG of AOI should cover the whole length of the desired strain direction. The selection of commonly used correlation criteria is of little concern in high-speed DIC processing on rock dynamic tests. The optimum subset size to achieve a good balance between accurate overall strain and high-resolution strain field with less noise is found close to three times of the largest speckle size. Coupled pre- and post-filtering are recommended to apply to correct the bias of overall strain at initial stage of loading and reducing the noise in strain field. A step size of 1 or 2 pixels is suggested for better identification of the strain localisation on rock surface. DIC-measured overall strain is valid until 50% residual peak stress at unloading stage with incremental correlation utilised at the peak stress. Highlights: VSG of AOI covering at least 43% length of specimen aligning with the desired strain orientation is reliable to extract the overall strain of rock. The subset size close to three times of the largest speckle size integrated with step size of 1 or 2 pixels is optimum in high-speed DIC measurement of rock strain. Coupled pre- and post-filtering in high-speed DIC is beneficial to achieve either accurate overall strain or good-quality strain field of rock. The overall strain of rock measured by high-speed DIC is valid to unloading stage at 50% residual peak stress with the incremental correlation. [ABSTRACT FROM AUTHOR]