Combination of discrete wavelet transform and ANFIS for post processing of spectroscopic signals

Abstract

We report on the capability of adaptive neuro fuzzy inference system (ANFIS) coupled with discrete wavelet transform (DWT) for the de-noising of absorption signals. The proposed method is examined for the de-noising of a CO2 spectrum which is experimentally traced in the near-infrared around 1.57 μm. This is performed by contaminating the spectrum with a various types of complicated noise figures produced by several approaches. It is found that DWT-assisted ANFIS is quite sufficient for CO2 signal recovery that is obscured by a huge amount of unknown noises without need of using a prior estimation of the noise feature. Eventually, enhanced signal-to-noise ratio of about 11.7 is obtained. © 2018, Springer Science+Business Media, LLC, part of Springer Nature.