Information Retrieval: Reconstructing Faces
Nigel W. Bond The Flinders University of South Australia
Kevin M. McConkey University of New South Wales
We all know that the face has a unique place in human affairs. We identify an enormous number of individuals on the basis of their face. Further, we read peoples' faces to determine their emotional state and to assist in our understanding of what they are saying to us1 ( Bruce, 1988). From work on the physiology of vision we know that there are special "face" detectors to be found in the right temporal lobe ( Perrett et al., 1985). However, we all know also that our face recognition system is prone to error. We see someone out of context--for example, the butcher at the beach--and we know that they are familiar, but we cannot remember who they are. Or, we see a friend and rush up to greet them, only to discover that the individual we are greeting is a perfect stranger. All of these facts are relevant to an understanding of face recognition and reconstruction. In what follows, we look at some of the problems of face recognition and reconstruction, and identify some solutions. Recent findings from the literature on face recognition and identification and the development of computer-aided procedures for the reconstruction of faces promise to revolutionize the way that police forces around the world deal with the complex problem of turning an eyewitness' account of a perpetrator's face into a lifelike picture of the original.
Let us begin by identifying some of the problems of face recognition and reconstruction. Bruce ( 1988) describes the case of Laszlo Virag who was con-____________________