Sparse spectral unmixing for activity estimation in γ-RAY spectrometry applied to environmental measurements

This paper presents a sparse spectral unmixing algorithm for activity estimation of radionuclides in γ-ray spectrometry. The spectral unmixing method aims to decompose a measured spectrum into spectral signatures of radionuclides, which is sensitive to the choice of the spectral signatures. The sparsity of the solution is imposed to identify the active radionuclides. Experimental results on simulated and real spectra show that the proposed method yields significant improvement for estimating radioactivity at low statistics. •New radioanuclides' activity estimation based on spectral unmixing.•Full spectrum and Poisson statistics of the detection process taken into account.•Reduction of the time to detect artificial radionuclides in aerosol filter samples.•Extended spectral unmixing to jointly identify the set of active radionuclides.