Algorithms and facial thermal imaging could identify drunks

Georgia Koukiou and Vassilis Anastassopoulos of the Electronics Laboratory at University of Patras, Greece, are developing software that can objectively determine whether a person has consumed an excessive amount of alcohol based on the relative temperature of different parts of the person’s face.

Writing in the International Journal Electronic Security and Digital Forensics, the team explained how such a system sidesteps subjective judgements based on behaviour, allowing law enforcement and other authorities to have what is described as ‘definitive evidence of inebriation’.

According to a statement, the team devised two algorithms that can determine whether a person has been drinking alcohol to excess based on infrared thermal imaging of the person’s face.

The first approach involves measuring pixel values of specific points on the person’s face, which are then compared with values in a database of scans of sober and inebriated people.

Given that alcohol causes dilation of blood vessels in the surface of the skin, hot spots on the face can be seen in the thermal-imaging scans, which can be classified as drunk or sober regions.