Reconstructing 3D faces in Cultural Heritance Applications

Abstract

Image-based 3D face reconstruction techniques can be used for generating a 3D model of a person’s face given a single or multiple images of the face of a person. Such techniques could be used for generating 3D models of faces appearing in cultural heritance artefacts, based on images captured with ordinary digital cameras. The ability to generate 3D reconstructions of faces appearing in cultural heritage artefacts will facilitate the development of numerous applications such as the generation of cultural heritage related 3D animations, short films, educational applications and customized educational games. In this paper we review key image-based 3D face reconstruction algorithms belonging to different categories and assess their applicability for reconstructing 3D faces appearing in cultural heritage artefacts. We also describe a dedicated 3D face reconstruction algorithm that can be used for efficient reconstruction of faces in cultural heritage applications based on a single image. The proposed algorithm requires the location of a number of landmarks on the face to be reconstructed. A deformable 3D shape model is used for generating an approximation of the 3D face geometry of the face shown on the artefact and texture mapping is used for generating the full 3D model. In the paper we present initial quantitative results and visual results obtained when the method was used for reconstructing 3D faces appearing in different types of cultural heritage artefacts such as statues, paintings, murals and mosaics.