Multiband imaging and characteristic spectrum extraction of optical fiber intrusion signal

Abstract

Abstract This paper proposes a multiband imaging method for optical fiber intrusion signals to realize the multi-dimensional analysis and the characteristic spectrum extraction. Firstly, the time-frequency analysis technique is performed on the initial collected intrusion signals to obtain the time-frequency distribution at each spatial intrusion position, on which energy band-wise integration is carried out along the frequency direction to generate spatial-temporal two-dimensional images in each frequency band. Then, these images are further stacked up as layers resulting in a spatial-temporal-frequency three-dimensional (3-D) cubic data. Based on this data, we extract the characteristic spectrums as the features of the intrusion signals. The real data experiments verify the effectiveness of this method.