A practical approach to optimizing the preparation of speckle patterns for digital-image correlation

The quality of strain measurements by digital image correlation (DIC) strongly depends on the quality of the pattern on the specimen\u2019s surface. An ideal pattern should be highly contrasted, stochastic, and isotropic. In addition, the speckle pattern should have an average size that exceeds the image pixel size by a factor of 3\u20135. (Smaller speckles cause poor contrast, and larger speckles cause poor spatial resolution.) Finally, the ideal pattern should have a limited scatter in terms of speckle sizes. The aims of this study were: (i) to define the ideal speckle size in relation to the specimen size and acquisition system; (ii) provide practical guidelines to identify the optimal settings of an airbrush gun, in order to produce a pattern that is as close as possible to the desired one while minimizing the scatter of speckle sizes. Patterns of different sizes were produced using two different airbrush guns with different settings of the four most influential factors (dilution, airflow setting, spraying distance, and air pressure). A full-factorial DOE strategy was implemented to explore the four factors at two levels each: 36 specimens were analyzed for each of the 16 combinations. The images were acquired using the digital cameras of a DIC system. The distribution of speckle sizes was analyzed to calculate the average speckle size and the standard deviation of the corresponding truncated Gaussian distribution. A mathematical model was built to enable prediction of the average speckle size in relation to the airbrush gun settings. We showed that it is possible to obtain a pattern with a highly controlled average and a limited scatter of speckle sizes, so as to match the ideal distribution of speckle sizes for DIC. Although the settings identified here apply only to the specific equipment being used, this method can be adapted to any airbrush to produce a desired speckle pattern