This paper reports human performance data from a series of psychophysi
cal experiments investigating the limits of stimulus parameters releva
nt to distinguishing a human face in a mug shot, In these experiments,
we use a two-alternative forced-choice paradigm for response elicitat
ion, We develop a benchmark that can be used to determine the performa
nce of a machine vision system for human face detection at different l
evels of image degradation. The benchmark is developed in terms of the
number of pixel blocks and the number of gray scales used in the imag
es, The paper presents a model of representation that can be useful fo
r recognition of faces in a database, and may be used to define the mi
nimum image quality required for retrieval of facial records at differ
ent confidence levels. Our results show that low-frequency information
in face images is useful since it is most resilient to degradation in
the image quality, The model is particularly relevant to the retrieva
l of facial images in large image databases, (C) 1995 Academic Press,
Inc.