Robust three-dimensional face reconstruction by one-shot structured light line pattern

Abstract Three-dimensional face reconstruction is important in many intelligent human-machine applications, e.g. automatic face authentication, human computer interface based entertainment and robotic surgery. Most state-of-the-art structured light methods require multiple projections of the designed patterns, which is not desirable for face reconstruction because it is uncomfortable or difficult for a person to keep his/her face rigid for a period of time. In this paper, we propose an approach to reconstruct the three-dimensional face by one-shot of structured light line pattern. The proposed one-shot 3D reconstruction approach calculates the distortion map from the distorted structured light line pattern. It relies on robust image processing methods and the generic features of human face to segment and cluster the projected structured light line pattern from one single captured image. The y coordinates of the clustered lines are extracted to form a low-resolution distortion matrix which is then transformed to full-resolution distortion map by spline interpolation. The 3D face is reconstructed from the distortion map and the calibrated parameters. Experimental results show that the proposed approach is able to reconstruct the three-dimensional human face robustly by one-shot.