Thermal image analysis for polygraph testing

TLDR

Periorbital blood flow is important in anxious states, as supported by prior laboratory findings. The method extracts facial temperature fluctuations via nonlinear heat‑transfer modeling, converts them to periorbital blood‑flow rates, and uses the slope of these rates as a feature for binary classification. We designed, developed, and tested a promising thermal‑image analysis method for polygraph testing, achieving an 84 % correct‑classification rate and potentially improving the reliability of traditional polygraph examinations.

Abstract

We have designed, developed, and tested a very promising thermal image analysis method for polygraph testing. The method achieved a correct classification rate of CCR= 84% on the test population to our avail. This method, once refined, can serve as an additional channel for increasing the reliability and accuracy of traditional polygraph examination. We extract subtle facial temperature fluctuation patterns through nonlinear heat transfer modeling. The modeling transforms raw thermal data to blood flow rate information. Then, we use the slope of the average periorbital blood flow rate as the feature of a binary classification scheme. The results come to support our previous laboratory findings about the importance of periorbital blood flow in anxious states.