Measuring human facial variation using 3D scanning and mapping

Description and assignment of individuals to specific populations is of significant interest to anthropologists, clinicians and forensic scientists. This research investigated the use of 3D facial images to create an unbiased, objective system for such tasks. Results were compared with each individual’s self-perception of ancestry and gender. Facial image acquisition was undertaken using a non-contact 3D surface scanner. The data were represented as an automatically generated dense set of landmarks which, with mathematical modeling, provided correspondence between homologous features for all the individuals in the study. Using this approach, the construction of population representatives was feasible. These archetypes were used to compare and depict areas of variation between homologous features of different groups allowing a quick analysis. Automated classification algorithms were derived from Principal Component Modeling. These were used to study the variation within the population around the archetype. Overall concordance between automated classification and declared ancestry and gender was 95% (195 of 206). No incorrect classification of an individual into a group that did not share at least one variable, either gender or self-perceived ancestry occurred. This analysis of3Dfacial images is a model for the visualization and description of other populations and the affinities between them.