Extraction and recognition of human body motion features based on pyroelectric infrared information

The infrared radiation of a walking person contains the information of walking gait and body shape. Pyroelectric infrared(PIR) sensors can detect human body motion within a fairly reasonable distance through its heat emissions. In this study, infrared radiation was detected by a PIR sensor with coded Fresnel lens arrays. Principal component analysis(PCA) method was employed to analyze the data obtained from different objects walking at different speeds for extracting their motion features. Then support vector machine(SVM) was used for classification and the highest probability of correct recognition was 85.38%. The study results show that it is feasible to employ PIR sensors to extract human motion features and to use the spectral data obtained from different objects walking at different speeds for classifi-cation. This proposed method also provides a new low-cost solution for monitoring low-security places.