Making faces with computers

Knowledge concerning the cognition involved in perceiving and remembering faces has informed the design of at least two generations of facial compositing technology. These systems allow a witness to work with a computer (and a police operator) in order to construct an image of a perpetrator. Research conducted with systems currently in use has suggested that basing the construction process on the witness recalling and verbally describing the face can be problematic. To overcome these problems and make better use of witness cognition, the latest systems use a combination of Principal Component Analysis (PCA) facial synthesis and an array-based interface. The present paper describes a preliminary study conducted to determine whether the use of an array-based interface really does make appropriate use of witness cognition and what issues need to be considered in the design of emerging compositing technology.