In suspect identification, witnesses examine photographs of known offenders in mug-shot albums.
The probability of correct identification deteriorates rapidly, however, as the number of mug-shots examined increases.
Feature approaches, where mug-shots are displayed in order of similarity to a witness's description of a suspect, attempt to increase identification success by reducing this number.
A methodology is proposed for the design, development, and evaluation of these systems based on computer simulations, experiments, and four classes of system performance measures : identification performance, retrieval rank, tolerance performance, and feature quality.
This approach was used to develop a suspect identification system for 640 mug-shots of known offenders.
System performance was good.
In three experimental tests,>90% of all witness searches resulted in target suspects retrieved in the first eight mug-shots.
The system is highly tolerant of witness errors and omissions.
Mots-clés Pascal : Identification, Individu, Reconnaissance image, Face, Assistance ordinateur, Criminologie, Homme, Suspect
Mots-clés Pascal anglais : Identification, Individual, Image recognition, Face, Computer aid, Criminology, Human
Notice produite par :
Inist-CNRS - Institut de l'Information Scientifique et Technique
Cote : 95-0532397
Code Inist : 002B30A10. Création : 01/03/1996.