Long-Distance Magnetostrictive Material-Based Distributed Magnetic Field Sensing by Modified Non-Local Means Denoising in OFDR

We present a long-distance magnetostrictive material-based distributed magnetic field sensing based on Modified Non-Local Means (MNLM) denoising in optical frequency domain reflectometry (OFDR). A 20 cm Terfenol-D magnetostrictive sheet is bonded at the location of single mode fiber (SMF) as fiber under test (FUT) over 500 m. Local Rayleigh scattering spectra measured using OFDR detects the strain coupled to the SMF induced by the magnetic field and magnetostrictive material. Since the sensing distance of OFDR is over 500 m, denoising is necessary. By searching for similar pixels using unweighted Euclidean distance, MNLM as an image denoising method effectively exploits the partial spatial similarity present in two-dimensional cross-correlation distribution of the local Rayleigh scattering spectra. Compared with traditional image denoising methods, including wavelet denoising (WD), Block-Matching 3D Filtering (BM3D) and Gaussian filtering (GF), only the proposed MNLM method can extract the local Rayleigh scattering spectra shift from noise under a sensing distance more than 500 m, a sensing spatial resolution of 7.5 cm and an average noise floor of 0.71 mT.