SPIE Membership Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
SPIE Photonics West 2019 | Call for Papers

2018 SPIE Optics + Photonics | Register Today



Print PageEmail PageView PDF

Defense & Security

Thermal analyzer enables improved lie detection in criminal-suspect interrogations

Real-time analysis of IR emission from eye and nose areas provides a nonintrusive method for evaluating statements.
15 February 2011, SPIE Newsroom. DOI: 10.1117/2.1201101.003452

The conventional polygraph, or lie detector, has long been used for questioning criminal suspects and even in some job interviews. The polygraph is based on a century-old system involving analysis of a person's involuntary psychophysiological responses to questions. Signals from sensors placed on the person's arm, fingers, chest, and abdomen detect changes in variables such as breathing and pulse rates, blood pressure, and perspiration. Information from the sensors was once recorded on a moving strip of paper but is now usually stored digitally. However, a major drawback of the standard polygraph test is its intrusive nature, which makes the subject anxious and uncomfortable and can, thus, affect the accuracy of the responses. An innocent but very nervous individual can fail a polygraph test. Conversely, the system can be defeated by a clever suspect. Therefore, development of noninvasive and more trustworthy lie-detection methods has become a highly desirable goal.

One of the more promising technologies for application to lie detection is thermal imaging. Since 2006, our team has developed and demonstrated1–5 a system we call the thermal analyzer for deception detection (TAD2). It analyzes far-IR data obtained remotely from a suspect's periorbital (around the eyes) and nostril areas during interrogation. With this data, a measured change in skin temperature from the two periorbital areas is converted to a relative blood-flow velocity. At the same time, the system deduces the person's respiration pattern from measured temperature changes around the nostrils (see Figure 1).

Figure 1. Far-IR skin-temperature (Ts) data from a subject's (a) periorbital and (b) nostril areas. The data reveals the person's relative blood-flow velocity (Vs) and breathing rate, respectively. dt: Time interval.

TAD2's only required instrumentation is a thermal-imaging camera that is sensitive to the far-IR spectrum (i.e., 7.5–13μm-wavelength bands) and a laptop computer. The camera is placed in front of the subject and oriented such that the person's head and part of the neck are in the image frame. The camera's elevation is a little lower than the subject's head and it is tilted upward so that two nasal passages are clearly visible. The camera is connected to the computer via a cable or wireless connection for real-time processing.

At the heart of TAD2 is a computer algorithm for automatic tracking of a subject's face position and movements. The algorithm can detect movement in a subject's face within 6.1ms and keep the face centered in a grid of x and y coordinates. The algorithm has an accuracy rate of almost 99%, enabling very precise monitoring of far-IR emission from the eye and nose areas.

TAD2 allows considerable flexibility for examiners. It supports three standardized examination methods, the modified general-question test (MGQT), the modified zone-comparison test (MZCT), and the relevant- and irrelevant-question test (I&R). With each of these methods—as with a conventional polygraph test—the questioner asks different kinds of questions: relevant, irrelevant, and control questions. Relevant questions pertain directly to the crime or other topic being investigated. Irrelevant questions have no bearing on the investigation. They are used simply to ease the subject into the examination. Control questions are ones that can be used for comparison purposes. For example, if a subject is suspected of having committed a burglary, the person might be asked, “Have you ever snuck into someone's house when they weren't home?”

Our baseline classification system for the scoring of answers to the relevant questions is based on a subject's responses to corresponding control questions. Scores of −1, +1, and 0 are applied to the answer to a relevant question if the recorded far-IR response is, respectively, stronger than, weaker than, or equal to that of the corresponding control question. A total score in the positive range or no lower than −2 suggests that the subject is giving truthful answers, while scores with greater negative values indicate lying. Figure 2 shows a screen shot of our current TAD2 setup, which provides a user-friendly interface embedded with six main operative sections, allowing the user to fill in the information of the subject, choose the appropriate examination method, and see the lie-detection result as soon as the examination has been completed.

Figure 2. Screen shot of our current TAD2 user interface.

We realized that tests conducted with volunteers might not yield the same kinds of results as examinations of actual crime suspects. To evaluate our system in real-world situations, we conducted a one-year test study in collaboration with Thailand's Department of Special Investigation. That test involved interrogation of 14 crime suspects and yielded a result of 84.4% accuracy. That is comparable to the results obtained in criminal interrogations with a conventional polygraph.

Obviously, we want to do better than that, and we think we can. Our immediate goal is to raise the accuracy of TAD2 to at least 90%. Our next step will be to develop more accurate baseline and scoring-adjustment methods through advanced statistics. We will also be refining our scoring methods with the use of neural networks—computer programs modeling the human brain—to delve into the thinking processes of interrogation subjects. Our future goals include getting TAD2 adopted by the Department of Special Investigation and other law-enforcement and homeland-security agencies in Thailand and elsewhere.

Sarun Sumriddetchkajorn, Armote Somboonkaew
National Electronics and Computer Technology Center
Pathumthani, Thailand

Sarun Sumriddetchkajorn is director of the Intelligent Devices and Systems Research Unit (IDSRU) and plays an important role in photonics research and development in Thailand. He is a senior member of SPIE, the Optical Society of America, and the IEEE Photonics Society, as well as a SPIE visiting lecturer.

Armote Somboonkaew is a researcher at the IDSRU's Photonics Technology Laboratory. His research areas include image processing, biosensors, and industrial opto-electronics.

1. S. Sumriddetchkajorn, A. Somboonkaew, Methods and systems for a non-contact analysis of psychophysiology for lie detection, Thai patent appl. 601002047, 2006.
2. S. Sumriddetchkajorn, A. Somboonkaew, Lie detection system and methods for real-time data tracking and analysis, Thai patent appl. 0701000585, 2007.
3. S. Sumriddetchkajorn, A. Somboonkaew, T. Sodsong, I. Promduang, N. Sumriddetchkajorn, A field test study of our non-invasive thermal image analyzer for deceptive detection, Proc. SPIE 6633, pp. 66331F, 2007. doi:10.1117/12.754220
4. S. Sumriddetchkajorn, A. Somboonkaew, Tracking improvement in our far infrared-based non invasive lie sensing system, Proc. Int'l Conf. Sens. (AsiaSense), pp. 183-187, 2009.
5. S. Sumriddetchkajorn, A. Somboonkaew, TAD2: the first truly non-intrusive lie detection system deployed in real crime cases, Proc. SPIE 7854, pp. 78540Z, 2010. doi:10.1117/12.886291