High-fidelity denoising for differential pulse-width pair brillouin optical time domain analyzer based on block-matching and 3D filtering

In the field of structural deformation and strain measurement for civil structural health monitoring, Brillouin optical time-domain analyzer (BOTDA) system with sub-meter high spatial resolution is required. Differential pulse-width pair Brillouin optical time-domain analyzer (DPP-BOTDA) is an excellent method to achieve the goal without the need to modify the system structure. But the signal-to-noise ratio (SNR) of DPP-BOTDA is poor due to the differential operation, which limits its practical applications. Here, we propose to use Block-Matching and 3D filtering (BM3D) algorithm to improve the SNR of DPP-BOTDA with less degradation of measurement accuracy and spatial resolution. The influence of each BM3D parameter on the denoising performance is fully analyzed in detail for the first time which helps the optimization of the algorithm. Three pulse pairs of 50/47 ns, 50/45 ns and 50/43 ns are used for the demonstration with a sensing range of 9.9 km. With the optimized BM3D, the experimental spatial resolution only degrades by 0.08 m, 0.05 m and 0.04 m after denoising when the SNR improvement is 9 dB, respectively. And the temperature uncertainty for the three cases is improved to be 0.79 degrees C, 0.36 degrees C and 0.24 degrees C, respectively. We believe that this research will improve the efficiency of the parameter optimization when using the BM3D method for noise reduction and contribute to the practical application of the DPP-BOTDA system.

Automated summary

In this study, we propose the use of the Block-Matching and 3D filtering (BM3D) algorithm to improve the signal-to-noise ratio (SNR) of the Differential pulse-width pair Brillouin optical time-domain analyzer (DPP-BOTDA) without significant degradation of measurement accuracy and spatial resolution. Our analysis of each BM3D parameter's influence on denoising performance helps optimize the algorithm. Experimental results show that with the optimized BM3D, the SNR of DPP-BOTDA is significantly improved, while maintaining high measurement accuracy for temperature sensing.