Spectral unmixing method for multi-pixel energy dispersive x-ray diffraction systems

摘要

An algorithm of spectral unmixing (SU) is presented, allowing the improvement of material classification accuracy based on the low spatial resolution images obtained by multi-pixel energy dispersive x-ray diffraction (EDXRD) systems. The method, which consists of signal subspace identification and endmember extraction, performs well even when the pixel count is rather small. Combined with SU, EDXRD has been utilized for liquid security screening for the first time. The spectra and abundance distributions of endmembers are extracted from the measured data sets corresponding to objects of different material composition, which demonstrates the validity of the proposed method. (C) 2017 Optical Society of America