Abstract. We investigate a number of approaches to pose invariant face recognition. Basically, the methods involve three sequential functions for capturing nonlinear manifolds of f...
A single face classifier has difficulty in detecting multiview faces under real and complex scenes due to various poses, cluttering environment and small size of faces. In this pa...
In this paper, we propose two novel methods for face recognition under arbitrary unknown lighting by using spherical harmonics illumination representation, which require only one t...
This paper describes a structure from motion and recognition paradigm for generating 3D models from 2D sets of images. In particular we consider the domain of architectural photog...
Anthony R. Dick, Philip H. S. Torr, Simon J. Ruffl...
—Human faces are remarkably similar in global properties, including size, aspect ratio, and location of main features, but can vary considerably in details across individuals, ge...