A generic approach for uncertainty quantification in stereovision DIC

Digital image correlation (DIC) is a wide-spread technique for measuring shape, motion and deformation in the field of experimental techniques. The complex optical-numerical nature of the technique however prohibits a straightforward and generic definition of error bars. Much uncertainty quantification has been done for 2D DIC, both experimental as numerical. Stereo-DIC is however much less investigated since the measurement chain becomes even more non-linear with the stereo-matching and the calibration stage. The novelty of this work lies in the introduction of an FE-based synthetic image generator, which is able to generate images as if they were obtained during a real experiment. Ground truth data becomes available by synthetically generating the images and individual experimental error sources can be extracted from the process. The validity and success of this technique is proven by the adoption by the stereo-DIC challenge, which is conducted under the auspices of the Society of Experimental Mechanics (SEM). Different sets of calibration and test images are generated for providing the DIC community a set of images for software testing, verification, benchmarking,... The target of this work thus is bringing more insight into stereo-DIC uncertainty quantification by programming a synthetic image generator and by investigating as many error sources as possible. This is reflected in the following contributions: • Development of a generic stereo-DIC image generator • Investigation of the calibration uncertainty with a Monte-Carlo approach • Uncertainty quantification of stereo-DIC by synthetic images • Investigation of relative camera motion, compensation methods and error estimationstatus: publishe