Research On Human Identification Authentication Based On Pyroelectric Infrared Information From Different Angles

Gait recognition is an emerging field of biometrics technologies, whose aim is to identify the individuals or detect physiological, pathological and psychological features based on their walking style. Therefore, this technique has a bright prospect and gains more and more attention from the field of biomedical information detection. The principle of gait identification based on pyroelectric infrared information is as follow, when a human walks, the motion of various components of the body, including the torso, arms, and legs, produces a characteristic signature, and infrared radiation is the only factor which influences the PIR's output. The features of human motion can be extracted by analyzing the output signals of PIR detector, then the recognition of different persons or different motion styles can be realized.Human body movement characteristic extraction for a walking person is investigated in this thesis, using a PIR sensor with Fresnel lens arrays. Temporal voltage signals collected from a PIR sensor were analyzed to extract the motion features of individuals for human identification. In this thesis, an infrared data acquisition system was first established. The signals from three detecting angles were collected simultaneously-the detecting angles were 60°, 90°, 120°.?The first way we used to extract feature was to extract spectrum information of the time-domain signal as the feature by using Fast Fourier Transform. The second way was to extract spectrum information of triple wavelet coefficient as the feature by using Fast Fourier Transform after Wavelet Transform. K-means cluster and BP network were utilized for human identification and authentication. What is more, the classification results of different methods were compared. Finally, we tried to fuse the features from different angles. The experiment result displayed that a better classification result was gained after feature fusion - the highest recognition rate was above 90%.The result of the thesis has shown that identification of human body by processing of the output signal of a PIR sensor can be achieved to some extent, and this was also low-cost human identification system programs in some low-security applications. At present, research on moving human identification based on pyroelectric infrared information is still in an initial stage, the results of this thesis will play an exploratory role in the development of this technology.