Automated photopeak detection and analysis in low resolution gamma-ray spectra for isotope identification

Automated isotope identification has long been an important problem in homeland security and nuclear emergency response. This process is difficult for low-resolution spectra because peaks can be significantly overlapping. Also, their areas can be difficult to determine because of the fluctuating baseline due to the Compton continuum across the whole spectrum. The wavelet transform stands out among many potential solutions of this problem, owing to its ability to de-noise noisy signals, pattern matching, and simultaneous multi-resolution signal analysis. In this paper, a novel wavelet-based algorithm for detecting peaks and measuring their areas is introduced, and specific wavelets are selected to find the optimal scale of signal analysis. Their abilities in locating peaks, resolving overlapping peaks, and determining peak areas are presented and assessed with both simulated signals and real gamma-ray spectra.