Skip navigation

Holistic facial composite systems: Implementation and evaluation

Holistic facial composite systems: Implementation and evaluation

Davis, Josh P. ORCID: 0000-0003-0017-7159, Gibson, Stuart J. and Solomon, Christopher J. (2017) Holistic facial composite systems: Implementation and evaluation. Face Processing: Systems, Disorders and Cultural Differences. Nova Science Publishers, pp. 41-54. ISBN 978-1536123986

[img] PDF (Author Draft Manuscript)
19288 DAVIS_Holistic_Facial_Composite_Systems_2017.pdf - Draft Version
Restricted to Registered users only

Download (507kB) | Request a copy


A facial composite is a likeness to a suspect’s face based on an eyewitness’s memory. Systems for constructing composites previously involved piecing together individual facial features. However, holistic (whole face), model based, composite construction has now become the most prevalent method used in the UK, as it more effectively matches the manner in which humans process faces. Here we describe the basic mechanism for generating holistic composite images and report on their effectiveness as a tool for locating suspects. Additional post-construction composite-enhancement methods such as morphing and perceptual stretch are also described. An accumulating body of empirical and field evidence has demonstrated that correct culprit identification rates are enhanced, and misidentification risks reduced, by the effective use of holistic composite systems in combination with post-construction techniques.

Item Type: Book Section
Uncontrolled Keywords: facial composite, holistic, evolutionary algorithm, eyewitness identification, morphing
Subjects: B Philosophy. Psychology. Religion > BF Psychology
Faculty / Department / Research Group: Faculty of Education & Health
Faculty of Education & Health > Applied Psychology Research Group
Faculty of Education & Health > Department of Psychology, Social Work & Counselling
Last Modified: 20 Feb 2018 16:56
Selected for GREAT 2016: None
Selected for GREAT 2017: None
Selected for GREAT 2018: None
Selected for GREAT 2019: None

Actions (login required)

View Item View Item


Downloads per month over past year

View more statistics