3D face reconstruction and verification using multi-view RGB-D data

The development of computer graphics and technologies leads to the emergence of different methods for presenting the three-dimensional information in the virtual world. In this paper we present an algorithm for 3D model reconstruction of a human face. A key aspect is that the face is captured from multiple views by a stationary low cost sensor providing both color and depth data. The different views are then registered into a common model. A verification method for assessing the accuracy of the reconstructed face is also presented.