Automatic Deceit Indication through Reliable Facial Expressions

The most broadly used tool for deceit measurement is the polygraph. Nonetheless, it is a limited method as it suffers from human operator subjectivity and the fact that subjects are aware of their measurement. This reduces performance because subject behavior changes, making their deceit or verity more difficult to measure, in addition to the fact that counter-measures can be planned in advance. The approach presented in this paper attempts to circumvent these problems by unobtrusively measuring facial expressions through computer vision analysis of standard color video. So called ´Reliable Expressions´ are leveraged as they are involuntary and have been thought to be able to be employed for deceit indication. The Deceit Indicators (DIs) themselves and algorithms to detect them were both tested.