Denoising of gamma-ray signals by interval-dependent thresholds of wavelet analysis

Fast Fourier transform and wavelet analysis methods were used to denoise electrical signals by a soft thresholding technique. A neutron pulse of 14 MeV is sent over a sample; as a consequence of the interaction between the sample and the neutrons, multi-spectral gamma rays are emitted by the sample. The gamma-rays were measured using three photo multiplier tubes, which detect optical signals coming from three filtered detectors made of plastic scintillator material. Applying wavelet analysis was possible to realize that the signal can be divided into three different regions. Each region has different thresholds; therefore, different frequency components can be used independently in each region. Comparisons of this method with the fast Fourier transform are presented. In this particular application, it was found that the wavelet transform produces a much better way of denoising the signals in terms of keeping the characteristic high frequency at the start of the signals; this feature allows the differential classification of the signals and the consequent identification of the component of the sample. The preliminary results presented here are the first attempt to identify the chemical composition of samples using this method.