A geometric morphometric evaluation of the Belanglo 'Angel' facial approximation

- The first evaluation of a forensic facial approximation's accuracy since 2001.
- Geometric morphometrics enables comparisons involving head pose variation.
- The results are that the facial approximation is accurate (p = 0.002).
- Although accuate, the inaccuracies would likely have disrupted recognition.

In August 2011, a 2D facial approximation was undertaken of remains discovered in Australia's Belanglo State Forrest; in October 2015, the young woman was identified. Referencing three photographs of the young woman as she appeared in life and a database of 64 sex, age, head pose and population matched images, the facial approximation is evaluated for relative shape accuracy through the application of geometric morphometrics. The results are that the facial approximation is significantly similar to the images of the young woman in facial morphology (p = 0.002) when most of the variance due to depicted head pose is removed from the analyses. The geometric morphometric analyses, however, also highlight the facial approximation's face and feature discrepancies, some of which would have likely disrupted familiar face recognition. Although predominantly verified methods were applied in 2011, they are limited in their predictive accuracy, not every feature of the face has a verified method to apply, and practitioner errors as well as photographic distortions are apparent. Furthermore, an assumption that the verified methods require inter-feature agreement (in this instance eye spacing and mouth width) was proved to be false. Overall this study shows that it is possible to assess the morphological accuracy of a forensic facial approximation when a number of antemortem images are available, though the influence of photographic distortion within 2D photographs will always preclude a precise metric assessment.