3D point clouds processing and precise surface reconstruction of the face

Sinusoidal fringe is projected to the face by a double fringe 3D scanner, and it is modulated by the face profile. Two separated point clouds of the face are obtained by the conversion of fringe image with demodulation and unwrapping processing. The system is calibrated by the calibration block. Automatic registration of two point clouds of the face is conducted by calculating the rotation matrix and translation vector with coefficient analysis. Triangle reconstruction patches of the face are obtained after the processing of the point clouds. NURBS face surface reconstruction with high precision is implemented by detecting curvature, generating rectangle meshes, and building UV parameter lines and surface reconstruction. Finally reconstruction error is analyzed. It shows that the obtained face surface meets the requirement of G1 continuity with a standard reconstruction deviation of 0.017604mm.