Multiscale Background Subtraction and Deconvolution Method for Weak-Peak Gamma Spectrum Under Low Peak-to-Compton Ratios

Although LaBr₃(Ce) detectors offer high detection efficiency for rapid in-situ measurement of low- and intermediate-level radioactive waste (LILW), their low energy resolution compromises the accuracy of conventional spectrum analysis methods, often resulting in the loss of weak peak information. To address this challenge, this research develops an adaptive Statistics-sensitive Nonlinear Iterative Peak-clipping (SNIP) parameter model, a multiscale Discrete Wavelet Transform (DWT)-based background correction model and a novel Skew-Normal distribution response function, and further proposes the multiscale WT-SNIP background subtraction and SN-ACTIVE deconvolution methods. Experimental and simulation results demonstrate that the WT-SNIP method accurately distinguishes peak and non-peak regions under low peak-to-Compton ratios (P/C) and adaptively performs effective background subtraction, yielding an average relative error of only 3.65% in net peak area, a 77.31% reduction compared with traditional SNIP, Gamma Vision and polynomial fitting. For overlapping peaks with area ratios of 1:1 to 1:10, WT-SNIP improved the peak-to-valley ratio (P/V) by 1.71–19.88 times, markedly surpassing the 1.03–1.60 times achieved by conventional methods. Furthermore, after precise background subtraction by WT-SNIP, SN-ACTIVE deconvolutes the broadened spectrum into an ideal incident spectrum, enabling more accurate weak peak search and area calculation, reducing errors by 30.00% and 75.98%, respectively, compared with the Gold algorithm, while improving convergence by three orders of magnitude. By integrating the proposed methods, adaptive multiscale background subtraction and spectrum deconvolution are achieved under low energy resolution, thereby enhancing the accuracy and robustness of qualitative and quantitative analyses in complex gamma spectrum. This research offers solid theoretical and technical support for the application of LaBr₃(Ce) detectors in rapid in-situ LILW measurements.