Background: Nuclear decay

electronic noise

statistic fluctuations

etc.

exist inherently. Therefore

the measured spectrum always has statistic fluctuation. Purpose: In order to reduce the statistical fluctuation and electronics noise in detector

translation invariant discrete wavelet transform (TI-DWT) wavelet modulus maxima spatial selectivity filter (TI-DWT-MSSNF) smoothing algorithm was put forward to preprocess data for the de-convolution of spectrum. Methods: The ?-spectrum was acquired by using ORTEC-8 channel ? spectrometer to measure the source numbered AMPU1103 (239Pu and 241Am) under vacuum conditions of ?0.03MPa. The ?-spectrum was obtained by using ? spectrometer and 137Cs source. Cubical smoothing algorithm with five-point approximation (“5-3”)

the traditional method of wavelet modulus maxima (WTMM) and TI-DWT-MSSNF were applied to smooth ? and ? spectra. Results: The study showed that TI-DWT-MSSNF method could eliminate statistical fluctuation more thoroughly

retain feature information better compared with “5-3” and WTMM. The D(r) values of TI-DWT-MSSNF were greater and ?2 values of TI-DWT-MSSNF were more close to 1 compared with those of “5-3” and WTMM. Conclusion: Comprehensive research indicates that it is feasible to reduce the statistical fluctuations of spectrum using TI-DWT-MSSNF. And TI-DWT-MSSNF outperforms both the “5-3” and WTMM. Abstract