Human and Nonhuman Recognition Using Pyroelectric Infrared Detector
- 296 Downloads
The auto regressive (AR) model of time series is utilized in this paper to recognize a human and nonhuman from pyroelectric infrared (PIR) signals. Through the wavelet transform, the signals are reconstructed by removing the noise from the original signals. The coefficients of the AR model are selected as the features for human and nonhuman recognition and calculated by the Burg algorithm. The classification experiments of a human and nonhuman are performed with a support vector machine. The recognition results for different PIR signals using the proposed AR-based features show high performance with an optimal recognition rate, which is up to 94.6 % and higher than that of the traditional time domain feature and transform domain method, such as the wavelet entropy and wavelet entropy of the double-density dual-tree complex wavelet transform.
KeywordsAutoregressive model Human and nonhuman recognition Pyroelectric infrared detector Support vector machine
Unable to display preview. Download preview PDF.
- 7.Cheng W.D., Dong Y.G.: Chin. J. Sci. Instr. 29, 1020 (2008)Google Scholar
- 8.Cheng X.: Research on Human Activity Recognition and Indoor Human Localization Based on Distribute Infrared Red Sensor Networks. Master’s Thesis, Shanghai Jiaotong University, Shanghai (2007)Google Scholar
- 9.Wang L.H., Gong W.G., Liu X.Y., Liang G.Q., He L.F.: Chin. J. Sci. Instrum. 12, 2487 (2009)Google Scholar
- 10.A.R. Kaushik, B.G. Celler, in Proceedings of the 28th Annual International Conference of the IEEE, vol. 8, New York, 2006, p. 5258Google Scholar
- 12.Kay S.M.: Model Spectral Estimation: Theory and Application. Prentice Hall, Englewood Cliffs (1988)Google Scholar