Detection of Deceit from Thermal Videos on Real Crime Database

Abstract

Detection of deceit is a very difficult task even for the experts. The time delay and requirement of experts for the existing polygraph test have led to requirement of noninvasive methods like thermal imaging. The major hurdle in the creation of a database of deceit is the simulation of guilt and lie in the participants. Most of the previous works use mock crimes with enacted responses of the participants. Here a novel experimental protocol is suggested where a real crime of stealing occurs. The recording has been done in a concealed manner. Also, the participants for the experiment are people with a strong tendency and record of habitual stealing. The experiment has been conducted at Medinipur medical college and hospital, India and has been approved by the ethical committee. The blood flow rate at the forehead and periorbital regions are found out from the thermal data. The rate of increase in blood flow rate shows that deceit could be detected from the thermal signatures.