| dc.description.abstract | Forensic facial reconstruction can assist identification by reconstructing a face of the
unknown person with the aim of its recognition by his/her family or friends. In the facial
reconstruction approach adopted in this study, a 3D average face template was digitally
warped onto a 3D scanned skull image. This study was carried out entirely on an Egyptian
population, and was the first of its kind.
Aims:
This study aimed to demonstrate that 3D facial reconstructions using the novel
methodology described could show significant resemblance to the faces corresponding to
the persons in question when they were alive. Moreover, using techniques previously
validated for facial reconstruction, the aim was to compare them to the method developed,
and to assess approaches used to determine the accuracy of 3D facial reconstructions.
Methods:
Initially, a pilot study was conducted using a database of laser scanned skulls and faces.
The faces were reconstructed using an average facial template generated by merging a
number of faces of similar population, sex, and age. The applicability, as well as the main
components of the facial reconstruction method, the single and average facial templates,
and the facial soft tissue thickness measurements, were investigated. Furthermore, in the
main study, the faces of computed tomography (CT) scanned heads of an Egyptian
population were reconstructed using average facial templates. The accuracy of the
reconstructed faces was assessed subjectively by face pool, and face resemblance tests,
and objectively by measuring the surface distances between the real and reconstructed
faces. In addition, a number of novel subjective and objective assessment methods were
developed. These included assessment of individual facial regions using subjective
resemblance scores, and objective surface distance comparisons. A new objective method,
craniofacial anthropometry, was developed by taking and comparing direct measurements
from the skull, and comparing the measurements from the real and reconstructed faces.
The studied cases were ranked according to all subjective, and objective, tests, and
statistically correlated.
Results and Conclusions:
The average facial templates showed a higher identification rate than the single face
templates. The approach of facial reconstruction used in this thesis showed a comparable
accuracy to many other facial reconstruction methods, yet was superior in terms of its
applicability, transferability, and ease of use. In the face pool tests, the younger assessors
were able to correctly identify the reconstructed faces better than older assessors.
Furthermore, the identification rate by the forensic anthropology experts was higher than
the non-experts. The former group showed the highest agreement between the observers
in giving the resemblance scores. Although there was a significant rank correlation
between the subjective and objective assessment tests, the subjective tests are influenced
by the assessors’ subjective characteristics (e.g., age, professional experience), thus
making objective assessment more reliable. However, in situations where subjective tests
are used, it is better to use the face resemblance tests and consult forensic anthropologists.
Also, Craniofacial Anthropometry, particularly the craniofacial angles, can successfully
indicate the accuracy of the facial reconstructions. Importantly, this study shows that
certain facial regions, particularly the cheek and the jaw, are more reliable than other
areas in the subjective and objective assessment of the facial reconstruction | en_US |
