The Doppelgänger effect? A comparative study of forensic facial depiction methods

Area of Science:

• Forensic Anthropology
• Facial Reconstruction
• Biometrics

Background:

• Facial depiction from skeletal remains is crucial in forensic identification.
• Reproducibility of 2D and 3D methods has not been consistently evaluated.

Purpose of the Study:

• To assess the reproducibility of 2D and 3D facial depiction methods from skeletal remains.
• To compare the accuracy and resemblance of different facial reconstruction techniques.

Main Methods:

• Blind study involving thirteen practitioners creating facial depictions from a skull model.
• Utilized 2D sketch, 3D manual, 3D automated, and 3D computer-assisted methods.
• Evaluated depictions using 3D geometric analysis, 2D face recognition algorithms, and familiar resemblance ratings.

Main Results:

• Five 3D depictions achieved ±2mm deviation for ≥50% of the face surface.
• No single 3D method consistently excelled across all evaluation metrics.
• 2D sketches yielded the highest-ranked matches in 2D face recognition comparisons.
• Familiar examiners rated most depictions (67%) as limited or moderate resemblance.

Conclusions:

• Consistent metric reliability can be expected with standardized facial depiction methods.
• Improved presentation standards for practitioners could enhance recognition in forensic cases.
• A combination of geometric accuracy and subjective resemblance is key for effective identification.