Evaluation of facial composites utilizing the EvoFIT software program

Abstract

Facial composites are traditionally created with the assistance of a sketch artist, and the resulting image is then circulated in the police force as well as the public community. However, with the advance of computer technologies and a better understanding of how facial composites are created, composite software systems have developed greatly. EvoFIT, an abbreviation for Evolutionary Facial Imaging Technique, is a computer program used to create composites based on the Darwinian concept. It allows a witness to select for global features of the face, that will in turn be combined together to create new faces that have a greater likeness to the offender. The EvoFIT program aims to boost the low recognition values of facial composite methods currently used. The purpose of this study is to evaluate production of two composites from the same person as a mechanism for improving performance. The use of a second composite, paired composites, and morphed composites is examined as mechanisms for boosting recognition. Ten sets of composites representing ten different volunteers (targets) were created using EvoFIT. The first composite in each set was named correctly 8.3% of the time, the second composites at 18.3%, the paired composites at 20%, and the morphed composites at 23.33%. The results support the theory that use of a second composite, a pair of composites, and morphed composites increases the number of instances in which namers correctly identify the target. This research suggests that it is valuable for a witness to construct a second composite using EvoFIT or similar software.