Removal of the continuum of X-ray spectra using morphologicaloperators

In energy dispersive X-ray fluorescence analysis, the estimation and removal of the continuum on which the X-ray spectrum is superimposed is a primary requirement. The algorithms commonly used are either complex, or in the case of, e.g., neural network algorithms, computation-intensive. They usually require strong constraints and/or hypotheses on the data or the shape of the continuum. Moreover, if the continuum amplitude is comparable to or bigger than the peak amplitudes, some of these algorithms can lose peaks. A new approach to continuum removal based on mathematical morphology is proposed here. The new algorithm permits fast continuum elimination without peak deterioration. Other than a rough estimate on the widths of the peaks, the new method does not require additional information about the spectrum. The method can also be applied without modification to background elimination from gamma ray spectra. This new method is described and results obtained from real and simulated spectra are discussed and evaluated