U.S. Tests $3.5 Million Computerized Lie Detector
The U.S. Department of Homeland Security will apportion $3.5 million to Rutgers University researchers to develop next-generation computerized lie detectors that can analyze the veracity of statements by studying subtle facial expressions, hand gestures, and other body language cues.
Rutgers computer science professor Dimitris Metaxas with the Center for Computational Biomedicine Imaging and Modeling says lie detector tests based on physiological indicators are about 50 percent reliable under even ideal conditions, but is convinced that expressions and gestures are not only harder to conceal, but are highly consistent regardless of race or culture.
Metaxas says this research is based on his earlier work, which focused on computer simulation of facial expressions and organ dynamics such as blood flow or heartbeat through image-based modeling.
Challenges highlighted by researchers include the incorporation of postures, gestures, and other new cues into the computer models, and determining the accuracy of movement captured under real-world conditions.
The Rutgers researchers have partnered with Lockheed Martin to develop and integrate 3D sensor technology for motion capture. The team expects their innovation will increase the efficiency of screening people at points of entry and border crossings, aid law enforcement in routine investigations, and improve security at embassies and other buildings.
Abstracted by the National Law Enforcement and Corrections Technology Center(NLECTC) from VNUNet (09/01/05); Jaques, Robert .