Detecting morphed passport photos: a training and individual differences approach

Area of Science:

• Cognitive Psychology
• Forensic Science
• Human-Computer Interaction

Background:

• Face photo verification is a common but error-prone method for identity checks.
• Identity fraud can exploit these errors, particularly with sophisticated techniques like morphed passport photos.
• Previous research highlights the vulnerability of face matching for unfamiliar individuals.

Purpose of the Study:

• To quantify the threat of morphed passport photos in identity fraud.
• To investigate the effectiveness of targeted training in detecting morphed images.
• To assess the correlation between morph detection ability and face matching accuracy.

Main Methods:

• Participants performed a morph detection task before and after receiving guidance.
• Two groups were formed: one received morph detection training, the other a control task.
• The Models Face Matching Test (MFMT) was administered to assess face matching skills.

Main Results:

• Baseline morph detection rates were initially poor across all participants.
• Morph detection training significantly enhanced the ability to identify manipulated images compared to basic guidance alone.
• Accuracy in identifying mismatches on the MFMT correlated positively with morph detection skills.

Conclusions:

• Targeted training is crucial for improving the detection of morphed identity documents.
• Enhanced morph detection abilities can contribute to more robust identity verification systems.
• Findings suggest potential countermeasures against morph-based identity fraud.