Synthesising a front-pose face image

This paper describes a method that calculates a 3D head model from a single 2D input face image using an hybrid image-model-based algorithm. A database of 3D head models, including 2D textures and 3D surfaces, is utilised for the construction of individual 3D heads from 2D face images. Once a 3D head is found, 2D face images can be rendered under different poses. By means of this method, a front-view face image is then rendered. Experimental results are reported. © 2004 IEEE.