The Compare Research And ASIC Implementation On Lie Detection Algorithms

Honesty is the base of the economical society, while intentional deception leads people into crime. Along with the development of lying detection recent years, the chemical index of brain has been one of the objects for lying detection research, meanwhile the lie detection micro-system becomes a trend.There are two objectives of this thesis. One is based on the advantages and disadvantages to analyze the known lying detection algorithms, to make the comparison between two new methods of lying detection. Another is to design the micro-system of lying detection based on the FPGA, according to the detail integrated requirement. Firstly, the known lying detection methods and its accuracies were analyzed for lie detection research paradigm learning. Aimed at the chemical index of brain and the auricular-brain reflection theory, I directly extracted the oxygen super-low spectrum from near-infrared otopoints in order to measure brain neurotransmitters indirectly. The key objects were to search novel features for lie or truth.Secondly, the near-infrared sensing technology was collected for sampling of the oxygen signals of auricular with PC system during card-lie detection tests. Two features were trained, one was K-means Cluster on SPSS software and the other was our super-low frequency auto-correlation algorithm.Finally, based on the chip of cyclone FPGA, I applied the method of super-low filtering and the near-infrared sensing circuit to achieve a quasi-integrated micro-system. Experiments'conclusions: the lie detection accuracies were 80% for super-low spectrum auto-correlation, 70% for K-means Cluster of SPSS software, and 80% for FPGA-based micro-system.This work will make a great contribution for the integration of lie detection ASIC.