Facial composite systems: review

摘要

In criminal investigation, a facial composite refers to a portrait of a sought unknown individual created by a forensic technician based on the memory of the eye-witness. Despite the use of computerised techniques in order to construct a composite image, until recently, the recognition rate on the basis of the offender’s portrait was very low. Although different software solutions were proposed, the main principles of the construction process remained the same since manual sketching. The process still relied on the detailed description of offenders and their individual features with the purpose of merging the individual features together in order to create a final portrait. Extensive psychological research, and inclusion of its results into the facial composite procedure has led to the development of “holistic” systems which enable much more efficient identification of individuals utilising facial composites. This paper compares the techniques used for the construction of composites, summarises psychologically—criminological research on human perception, processing and recall of unknown faces, as well as a review of the new generation systems which automatically generate facial images in terms of facial representation and the applied optimisation method (GA).