Adaptive Block-Matching and 3D Denoising for Φ-OTDR Under Ultra-Low SNR Conditions
Jingming Zhang, Yaxi Yan, Shuaiqi Liu, Xingwei Chen, Alan Pak Tao Lau, Changyuan Yu, Liyang Shao
Abstract
This study presents a novel approach for extracting perturbations in phase-sensitive optical time domain reflectometry (Φ-OTDR) under challenging ultra-low signal-to-noise ratio (SNR) conditions. To effectively enhance the SNR of Φ-OTDR signals, a block-matching and 3-D (BM3D) based denoising scheme was proposed. 1-D Φ-OTDR traces are converted to 2-D images that encapsulate both temporal and spatial information, enabling the utilization of the data correlation in both domains through BM3D to reduce the signal noise. This approach effectively preserves intricate details of Rayleigh signals while mitigating the impact of noise. Moreover, an adaptive parameter selection method tailored to the characteristics of reflected backscattered signals is developed to facilitate a more convenient and robust implementation of the denoising algorithm. Experimental validation is accomplished by performing distributed vibration sensing over an 80 km optical fiber with a spatial resolution of 10 m. Results show a substantial enhancement in SNR from 1.77 dB to 40.6 dB at the end of the test fiber. We firmly believe that the proposed approach holds significant potential for widespread application in long-distance distributed sensing scenarios operating under low SNR conditions.