3DMM for accurate reconstruction of depth data

Abstract

In this paper, we propose a framework to derive accurate reconstructions of the 3D face surface from low resolution depth frames by means of a 3D Morphable Model (3DMM). By using a 3DMM specifically designed to support local and expression-related deformations of the face, we propose a two-steps 3DMM fitting solution: initially the model is warped based on landmarks correspondences; subsequently, it is iteratively refined by means of a mean-square optimization on the nearest-neighboring vertices. Preliminary results show that the proposed solution is able to derive faithful 3D models of the face, both for low-and high-resolution scans; quantitative results also evidence the higher accuracy of our approach with respect to methods that use one step fitting based on landmarks. In addition, we employed the 3DMM fitting to learn expressions specific coefficients, that can be further applied to the deformed models so as to generate subject-specific expressive scans, while the fitting procedure allows maintaining unaltered the general surface topology of the original scans.