The Characteristic Extraction Method of Fiber Intrusion Signals Based on Band Cutting and Imaging

Abstract

This thesis put forward a band cutting and imaging to extract characteristic for optical fiber intrusion signals. Firstly, frequency-time distribution of every dimensional intrusion location is obtained by the frequency-time analysis technique of original collected signals. Then, energy band integral goes along the frequency direction. The 2-D images are further stacked to form layers, leading to 3-D space-time-frequency cubic data. Lastly, the corresponding time and space feature spectrum based on the 3-D cube data is as the feature of the intrusion signals. The actual data experiments show that multi-band cutting and imaging is a very effective method for feature extraction and type identification of fiber-optic intrusion signals.