Facial biometrics could help stop or convict some serial killers: researchers

facial-recognition-database

In a development as grim as any aspect of biometrics, researchers believe they have found a way to potentially link some missing persons and murder victims to some serial killers by using facial features.

It is generally accepted in forensics circles that at least some sexually motivated serial killers make victims of people who share some facial features and geometries. Based on this premise, three Australian researchers decided to see if they could find biometric links among a serial killer’s victims.

Their results, based on photographs of three female victims and one female control, indicated a qualified yes.

The victims were preyed upon by the monstrously prolific murderer Ted Bundy. The fourth image was pulled at random from the Yale Face Databased B, 2019.

The Murdoch University (Perth) researchers coined the term face similarity linkage for their technique in a paper published in the journal Expert Systems.

Seventeen points of comparison were identified and applied to each victim’s image. No statistically meaningful differences were found among that group. However, scientists spotted 11 facial measurements of the victims that differed significantly compared to the control subject.

Analysis was a “hand-measured extension of facial photo-anthropometry,” and AI would be required to continue the work at larger scales.

They note that facial recognition is significantly dependent on the quality and condition of images, limiting their technique’s success rate. In fact, “caution should be exercised when using face similarity linkage.”

It should only be used for two purposes at this stage, according to the researchers. First, to help exclude assumptions, hunches and theories and, second, as an intelligence tool to better apply investigation resources.

Having said that, the research indicates that face similarity linkage could play a coordinating role with DNA phenotyping. It could shrink the pool of potential candidates generated by phenotyping. It might be able to suggest a face shape, for instance.

Related Posts

Article Topics

 |   |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Read This Week

Featured Company

Biometrics Research

Biometrics White Papers

Biometrics Events

Explaining Biometrics