Measuring displacement and strain fields at low observable scales of complex microstructures still remains a challenge in experimental mechanics often because of the combination of low definition images with poor texture at this scale. This is the case for cellular materials, for which complex local phenomena can occur. The aim of this paper is to design and validate numerically and experimentally a Digital Image Correlation (DIC) technique for the measurement of local displacement fields of samples with complex cellular geometries (i.e samples presenting multiple random holes). It consists of a DIC method assisted with a physically sound weak regularization using an elastic B-spline image-based model. This technique introduces a separation of scales above which DIC is dominant and below which it is assisted with image-based modeling. Several in-silico experimentations are performed in order to finely analyze the influence of the introduced regularization lengths for different input mechanical behaviors (elastic, elasto-plastic and geometrically non-linear) and in comparison with true error quantification. We show that the method can estimate complex local displacement and strain fields with speckle-free low definition images, even in non-linear regimes such as local buckling or plasticity. Finally, an experimental validation is proposed in 2D-DIC to allow for the comparison of the proposed method on low resolution speckle-free images with a classic DIC on speckled high resolution images.