A novel technique for face recognition using range imaging

Abstract

We consider a novel technique for recognizing people from range images (RI) of their faces. Range images have the advantage of capturing the shape variation irrespective of illumination variabilities. We describe a procedure for generating RI effaces using data from a 3D scanner, and registering them in the image plane by aligning salient facial features. For statistical analysis of RI, we use standard projections such as PCA and ICA, and then impose probability models on the coefficients. An experiment describing recognition effaces using the FSU 3D face database is presented.