Performance comparison of various deconvolution algorithms realizing high spatial resolution Brillouin optical time domain reflectometer

Applying the image deconvolution algorithm to Brillouin optical time domain reflectometer (BOTDR) can enhance the spatial resolution. In this paper, the blurring effect of point spread function on Brillouin frequency shift of sensing fiber under different pump pulses is numerically simulated based on the two-dimensional convolution theory of BOTDR. The recovery effects of several classical image deconvolution techniques, including Wiener filtering, regularized filtering, Richardson- Lucy method and blind deconvolution, in improving the spatial resolution of BOTDR are experimentally compared. The experimental results show that all four filtering methods can improve the spatial resolution from 2.5 m to 20 cm. In this application, the Wiener filtering, regularized filtering, and the Richardson- Lucy method provide almost the same restoration, while the blind deconvolution method provides a poorer restoration.