Creation of reliable forensic databases for holistic face construction

Perception, Cognition and Neuroscience (PCN) Research Group

As part of an international collaboration between the University of Central Lancashire and Luleå University of Technology, this project is developing appropriate methods for creating face databases and procedures for effective construction of holistic face composites.

Research 

Construction of identifiable facial composite images is vital for effective policing. Considerable research has indicated the importance of recognition type systems for constructing facial images from eyewitness memory (Frowd, 2021). We have been developing the EvoFIT facial-composite system for over 20 years, shaping each stage of the process from initial interview, composite system to methods for enhancing recognition (e.g., Brown et al., 2019, 2020; Fodarella et al., 2017; Frowd et al., 2006, 2010, 2012, 2015; Martin et al., 2018; Skelton et al., 2019).

Given the important forensic need for effective databases for holistic composite systems (e.g., Frowd et al., 2006; Richardson et al., 2020), the current project explores how best these models should be created and formally assessed for reliability. The project focuses on construction of white Caucasian faces in the UK and Sweden.

Primary goals and objectives

  • Reliability analysis for construction of holistic face databases
  • Formal description of procedures for forensic image creation
  • Enhancement techniques for best forensic practice

Impact

The project provides a framework of forensic procedures, a ‘gold’ standard, for production of effective models for holistic face construction.