Application of Gaussian curve fitting method in gamma energy spectrum overlapping peak identification

Due to the influence of the spectrum analysis detector resolution, the interference peaks between the analysed samples, the overlapping peaks in the spectrum analysis is very common. The appearance of overlapping peaks seriously affects the accurate identification of the analysed samples, therefore, the accurate identification of overlapping peaks is very necessary in the field of gamma energy spectrum. In this paper, a novel method for identifying the peak parameters of overlapping peaks based on Gaussian curve fitting is proposed. The method uses a second-order derivative of the Gaussian function convolution algorithm to obtain the initial peak parameters for the curve fitting. This step reduces the curve fitting time and improves the fitting accuracy. Subsequently, the values of the overlapping peak parameters (including peak height, peak width, peak centre position) with different step transformations were taken as fitting parameters for each fit. When the fitting error between the fitted peaks and the original data was the smallest, the values of the peak parameters were output. The simulation results show that the peak parameters can be identified properly even with a low signal-to-noise ratio. The simulated overlapping peak experiments with different separations demonstrate that the overlapping peak recognition results of this method are better than those of the wavelet transform method and particle swarm algorithms. Finally, the measured spectra demonstrate the feasibility of the novel method in gamma energy spectrum overlapping peak identification.