Interoperator test for anatomical annotation of earprints.

Area of Science:

• Forensic Science
• Biometrics
• Pattern Evidence Analysis

Background:

• The Forensic Ear Identification (FearID) project aims to quantify evidence strength from crime scene earprints.
• A large database of over 1200 earprint donors has been established.
• Current methods rely on expert anatomical annotations of minutiae and landmarks.

Purpose of the Study:

• To validate the stability and objectivity of the manual earprint annotation process.
• To assess interoperator variability in earprint analysis.
• To identify limitations in current knowledge-based approaches for earprint comparison.

Main Methods:

• A knowledge-based approach using expert anatomical annotations.
• Comparison of earprint pairs using Vector Template Matching (VTM).
• A validation experiment to study the stability of the annotation process.

Main Results:

• Significant interoperator effects were observed in earprint annotation.
• Individual operators showed more consistent results than comparisons between different operators.
• Variations in annotation point selection and clicking frequency indicate subjectivity.

Conclusions:

• Manual annotation processes in earprint comparison are operator-dependent and may not be optimal for objectifying results.
• The findings are relevant to other forensic identification disciplines relying on manual analysis, such as fingerprints and handwriting.
• Further research is needed to develop and implement more objective and automated methods for forensic identification.