作者: Uday Jain , Bozhao Tan , Qi Li
DOI: 10.1109/ICASSP.2012.6288219
关键词: Computer science 、 Artificial intelligence 、 Facial recognition system 、 Face detection 、 Pattern recognition (psychology) 、 Computer vision 、 Pattern recognition 、 Face (geometry) 、 Word error rate 、 Feature extraction
摘要: In this paper, we present a non-intrusive lie detection system based on thermal imaging technologies. The consists of the following modules: camera, face and tracking, landmark detection, feature extraction, pattern recognition for concealed knowledge inference. We have discovered most sensitive areas human to monitor facial temperature changes. Detection algorithms are then developed identify from automatically. Face tracking is used directly video images detect regions interest (ROI) extract features achieved an equal error rate (EER) 16.5% in 16 subjects test data. Our non-contact method using data achieves similar or better accuracy as traditional intrusive methods, such polygraph EEG.