Abstract: For both 2D and 3D measurements by correlation techniques, speckle patterns can be composed of rigid particles (RP). In this paper, a study of the influence of these RP on digital image correlation (DIC) strain error assessment is performed experimentally and numerically. Firstly, it is shown that prior to the influence of pattern, it is necessary to have an appropriate calculation strategy of initial values for DIC iterative process. Then for simulated images, the RP pattern presents a similar behaviour as the one of a deformable pattern to which a noise is added. However, for experimental images, this noise is less predominant on error assessment than the one because of deteriorations of patterns for large strains.