Accelerated Fast BOTDA Assisted by Compressed Sensing and Image Denoising

Abstract

We propose and experimentally demonstrate a scheme for accelerated fast BOTDA. The effect of signal-to-noise ratio (SNR) on recovery performance of compressed sensing is simulated and analyzed, it is found that a reduction in SNR requires much larger frequency data to recover the original Brillouin gain spectrum (BGS). To enable a high recovery probability, Block-Matching and 3D filtering (BM3D) algorithm is employed to enhance the SNR of Brillouin time trace and reduce the number of averages. Combining with a principal component analysis (PCA) based compressed sensing technique, the Brillouin gain spectrum (BGS) can be successfully reconstructed from only 37.5% frequency data. In the experiment, 75 randomly selected frequency data is acquired to reconstruct the BGS. Distributed strain sensing is achieved over 15 km single-mode fiber with 3 m spatial resolution and 0.52 MHz Brillouin frequency shift (BFS) uncertainty. Due to the accelerated process, the measurement time with 40 averages is less than 0.5 s.