The Application Of Digital Image Processing Technology On Brillouin Optical Time Domain Analyzer

Distributed optical fiber sensing(DOFS)systems have attracted the attention of both the academic and industrial sectors,owing to a series of advantages over conventional sensors.Typically,Brillouin optical time domain analyzer(BOTDA)based Brillouin gain spectrum(BGS)measurement has been widely utilized in safety and health monitoring of large infrastructures,for its significant superiority of ultra-long sensing distance,metric spatial resolution and high measurement precision.It has become new research hotspots that how to further enhance the performance of BOTDA sensor.Due to the similarity of DOFS signal,digital image denoising could be an impactful way to remove the noise,and we focus on applying digital image processing technology to enhance the performance of BOTDA sensing signal in this paper.By numerical and experimental study,we compare the noise level estimation of three different methods for BOTDA: calculating the standard deviation(STD)of the measurements,wavelet-based estimation algorithm,and patch-based estimation algorithm proposed in this paper,which selects weak textured patches of BOTDA signal and then estimates noise level using principal component analysis(W-PCA).The results shows: W-PCA and the mean of STD can accurately estimate the noise level,while wavelet-based estimation algorithm overestimates the noise level.Nevertheless,the STD estimation value has huge fluctuation along the length,while the W-PCA is relatively robust for its global consideration.Secondly,for the acquisition of key parameter(noise level)of non-local means(NLM)image denoising algorithm,we proposed NLM based on noise level estimation,and it further improves the engineering practicability of the NLM algorithm.Then,the plain BOTDA experimental data denoising processing shows that NLM based on W-PCA effectively suppresses the noise while keeping the detail of perturbation,but the NLM with overestimated noise(wavelet-based estimation algorithm)induces signal distortion.Finally,we make a comprehensive study of noise level estimation and NLM denoising for advanced BOTDA for the first time.A 157.71 km advanced BOTDA is been demonstrated,assisting by hybrid distributed amplification and optical pulse coding.Optimal NLM is utilized to process experimental data of BOTDA,where the values of W-PCA,STD and filter-based estimation respectively are as the input of NLM.The denoising processing significantly enhances the SNR of the data,and only the detailed texture of the NLM based W-PCA denoising signal is well kept(8m spatial resolution and 10°C).With the help of W-PCA,NLM algorithm reduces the measuring uncertainty from ±1.1MHz down to ±0.65MHz(69% increase in figure of merit).For other DOFS systems,we can use the W-PCA to estimate the noise level of signal,and then the NLM can also improve the performance of such sensing systems.