This work concerns the shape identification of curvilinear objects, for example bent beams or wires in mechanics. The beam’s digital picture is analyzed with the introduced Virtual Image Correlation method. This one consists in finding the optimal correlation between the beam’s image and a virtual beam, whose curvature field is described by a truncated series. The gray level and amplitude of the virtual beam does not need to reproduce exactly the ones of the physical beam image. The analytical form of the optimal shape allows one to derive mechanical properties: the identification of the Young’s modulus of a bar is given as an example. We will also show the robustness of the method with regards to the quality of the image.