Wavelet Approach for Operational Gamma Spectral Peak Detection - Preliminary Assessment
Gamma spectroscopy for radionuclide identifications typically involves locating spectral peaks and matching the spectral peaks with known nuclides in the knowledge base or database. Wavelet analysis, due to its ability for fitting localized features, offers the potential for automatic detection of spectral peaks. Past studies of wavelet technologies for gamma spectra analysis essentially focused on direct fitting of raw gamma spectra. Although most of those studies demonstrated the potentials of peak detection using wavelets, they often failed to produce new benefits to operational adaptations for radiological surveys. This work presents a different approach with the operational objective being to detect only the nuclides that do not exist in the environment (anomalous nuclides). With this operational objective, the raw-count spectrum collected by a detector is first converted to a count-rate spectrum and is then followed by background subtraction prior to wavelet analysis. The experimental results suggest that this preprocess is independent of detector type and background radiation, and is capable of improving the peak detection rates using wavelets. This process broadens the doors for a practical adaptation of wavelet technologies for gamma spectral surveying devices.