A Digital Image Correlation Algorithm with Light Reflection Compensation

This work focuses on the influence of light reflection on Digital Images Correlation results at the macroscopic scale, and on a way to circumvent this problem. It shows that the local displacement uncertainty rises up to 5 times the usual one when a reflection occurs. An observation of the topography of the speckle reveals an important sub-pixel roughness that explains the grey level fluctuations at the pixel scale, spoiling the calculation of the gradient of the texture. To circumvent this problem, a new DIC algorithm is proposed, based on a single minimization with several pairs of images where the reflections are located in different regions. For each image, weighting functions are used to ‘exclude’ the reflection regions from the calculation, while the necessary information is obtained from the other images.