Towards Criteria Characterizing the Metrological Performance of Full-field Measurement Techniques: Application to the Comparison Between Local and Global Versions of DIC

International audienceUsers of full-field measurement methods like Digital Image Correlation (DIC) often aim to perform measurements with the best trade-off between spatial resolution, bias and measurement resolution. Whenever two full-field methods are compared, it is essential that these criteria are taken into consideration. Recently a metrological efficiency indicator for full-field measurements has been proposed and discussed. This indicator combines measurement resolution and spatial resolution. It has been shown to be invariant to the subset size in the case of Local DIC. The goal of this article is to discuss a method, which determines both the spatial and the measurement resolutions for a given bias for two different DIC methods, in order to obtain the metrological efficiency indicator for each of these methods. The benefit of this indicator is that it does not depend on setting parameters such as the subset size, which are chosen by the user. As such, it can be considered as intrinsic to each technique, thus enabling fair comparison. Local DIC and triangular finite element based Global DIC will be the subject of this investigation. With this setting, their respective subset and triangular element sizes will be related to the spatial resolution of both methods for a given acceptable bias. By using the metrological efficiency indicator, the performance of the two methods will be compared and discussed to a new level of detail. Generally speaking, the indicator shows that the metrological performance of both methods is similar, confirming their popularity. However, it will be shown that, depending on the choice of what an acceptable bias is, one of the method may be preferred to another. The results show that for the specific DIC versions used in the study, for cases for which a significant bias is acceptable, Local DIC outperforms Global DIC, while the opposite is true in the case for which the bias requirements are more stringent. Finally, the quadratic versions of both DIC versions are shown to significantly outperform their respective linear versions