The understanding and the prevention of damage mechanisms, which have an impact on the hydrocarbon production and recovery rates, are of paramount interest for reservoir engineers. The modelling of such coupled processes relies essentially on experimentally obtained data, which characterize the macroscopic mechanical and transport properties. This approach however cannot account for the multi-scale structural heterogeneities of the considered rocks, in spite of their fundamental importance. The microstructural characterization of damage is usually based on ‘post-mortem’ observations of the samples, which provide both qualitative and quantitative information about the effects of the mechanisms activated at the grain scale and at intermediate scales, at a pervasive stage of damage after sample unloading. New techniques provide more quantitative and direct methods to follow the deformation history and the eventual development of localization and damage. In this study, the 2D Digital Image Correlation (DIC) technique has been applied to sequences of images taken from carbonate samples during uniaxial compression tests. Several scales have been considered, ranging from the centimetric scale of the samples to the local scale of their microstructure. For this purpose both optical observations and Scanning Electron Microscopy (SEM) were used. Although the macroscopic strain at failure was very small (< 0.2%), the DIC technique has proven to be reliable, provided one selects carefully image acquisition conditions and DIC parameters, as highlighted in our discussion on the uncertainties and the evaluation of errors. This technique has allowed us to quantify both the global and local strain fields during the deformation process. We have thus been able to precisely identify the localizations of damage and the local compaction mechanisms, and to relate them to the characteristic structural heterogeneities of the tested carbonate.