Segmented Noise Reduction Based on Brillouin-Spectrum-Partition in Brillouin Optical Time Domain Sensors

In this paper, the necessity of segmented image denoising in the signal post-processing of Brillouin Optical Time Domain sensors (BOTDS) is investigated firstly, then a novel cross-correlation based Brillouin-spectrum-partition (BSP) method is proposed to denoise the noisy Brillouin gain or loss spectrum in BOTDS. In the BSP filter, the non-correlation factors of adjacent Brillouin scattering spectrum are firstly computed, then the Brillouin spectrum along the fiber is divided into transition sub-images with large changes in Brillouin frequency shift (BFS) and non-transition sub-images with small changes in BFS, and finally the transition and non-transition sub-images are filtered with linear and non-linear filters, respectively. By using an 800m-length G657 sensing fiber with five temperature events in a proof-of-concept experiment, a BFS measurement accuracy improvement of 22% in non-transition section, and a spatial resolution (SR) enhancement of 84% in transition Section can be achieved by using the proposed BSP filter, compared to the results obtained by using non-local means (NLM) filter. Besides, a BFS measurement accuracy improvement of 54% in non-transition section, and a SR enhancement of 71% in transition section, can be achieved compared with the results obtained by using Gaussian filter. Moreover, to deal with the noisy Brillouin spectrum of the 800m-length fiber by using the proposed BSP filter, we obtain a processing time reduction of 37% compared with that by using NLM filter.