Forensic Facial Comparison: Current Status, Limitations, and Future Directions

Abstract

Global escalation of crime has necessitated the use of digital imagery to aid the identification of perpetrators. Forensic facial comparison (FFC) is increasingly employed, often relying on poor-quality images. In the absence of standardized criteria, especially in terms of video recordings, verification of the methodology is needed. This paper addresses aspects of FFC, discussing relevant terminology, investigating the validity and reliability of the FISWG morphological feature list using a new South African database, and advising on standards for CCTV equipment. Suboptimal conditions, including poor resolution, unfavorable angle of incidence, color, and lighting, affected the accuracy of FFC. Morphological analysis of photographs, standard CCTV, and eye-level CCTV showed improved performance in a strict iteration analysis, but not when using analogue CCTV images. Therefore, both strict and lenient iterations should be conducted, but FFC must be abandoned when a strict iteration performs worse than a lenient one. This threshold ought to be applied to the specific CCTV equipment to determine its utility. Chance-corrected accuracy was the most representative measure of accuracy, as opposed to the commonly used hit rate. While the use of automated systems is increasing, trained human observer-based morphological analysis, using the FISWG feature list and an Analysis, Comparison, Evaluation, and Verification (ACE-V) approach, should be the primary method of facial comparison.

Keywords: CCTV; FISWG; disguises; face mapping; facial identification; forensic facial comparison; human identification; morphological analysis; photography.

Figure 1

Example of a forensic facial comparison analysis process between a wildtype (WT) photograph and a standardized (ST) photograph from the Wits Face Database [41] sample images in the SAPS court chart format. The individual facial features are numbered, analyzed, compared, and evaluated between the two images using the FISWG feature list [47]. Features marked in blue indicate morphological similarity between the two images, while features marked in red indicate morphological dissimilarity. In the example provided, skin color appears different due to lighting discrepancies in the two images (red 1); however, skin texture appears similar (blue 1). The facial images used for Figure 1 are images of the corresponding author of the present manuscript and are part of the sample images of the Wits Face Database [41], reproducible under an open access license distributed under the terms of the Creative Commons Attribution License. This license permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The images can be found in the Wits Face Database data note, including the supplementary material for the Wits Face Database [41].

Figure 2

Visual summary of the validation studies testing morphological analysis across realistic photographic and CCTV conditions [21,48,49] using sample photographs and CCTV stills from the Wits Face Database [41]. Images (A) to (F) are samples of the target images from each set of conditions analyzed that were compared to the central image arising from the standardized photographs captured for each participant. All major statistical results and the details of the conditions of each comparison cohort are presented. Representative images of each condition are arranged from A to F in a clockwise order according to descending chance-corrected accuracy. The conditions of analysis were as follows: wildtype informal photographs (A) of similar quality to the standardized photographs; eye level digital CCTV still images (B); standard digital CCTV still images (D) with sunglasses (C) and with brimmed caps (E); and monochrome analogue CCTV still images (F). Key: CCA = chance corrected accuracy; FPR = false positive rate; FNR = false negative rate; OA = observer agreement; RES = resolution; SCD = subject-to-camera distance; AOI = angle of incidence; N = number of comparisons. The facial images used for Figure 2 are images of the corresponding author of the present manuscript and are part of the sample images of the Wits Face Database [41], reproducible under an open access license distributed under the terms of the Creative Commons Attribution License. This license permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The images can be found in the Wits Face Database data note, including the supplementary material for the Wits Face Database [41].

Figure 3

Flow diagram of the recommended morphological analysis process. This approach to morphological analysis uses an ACE-V method in conjunction with the FISWG feature list [47], with the inclusion of the ENFSI’s image quality triaging [114] and the use of the South African Police Services (SAPS) scoring criteria [17] as adapted for research application [21]. Statistical analyses for research use are also recommended based on our recent work [48] to allow for more detailed result interpretation and comparison among future studies.