guest ::
| Home > Publications database > Automatic reduction of large x-ray fluorescence data-sets applied to XAS and mapping experiments |
| Home > Publications database > Automatic reduction of large x-ray fluorescence data-sets applied to XAS and mapping experiments |
| Book/Report/Dissertation / PhD Thesis | PUBDB-2017-00957 |
; ; ; ; ; ; ;
2017
Verlag Deutsches Elektronen-Synchrotron
Hamburg
This record in other databases:
Please use a persistent id in citations: doi:10.3204/PUBDB-2017-00957
Report No.: DESY-THESIS-2017-007
Abstract: In this thesis two automatic methods for the reduction of large fluorescence data sets are presented.The first method is proposed in the framework of BioXAS experiments. The challenge of this experiment is to deal with samples in ultra dilute concentrations where the signal-to-background ratio is low. The experiment is performed in fluorescence mode x-ray absorption spectroscopy with a 100 pixel high-purity Ge detector. The first step consists on reducing 100 fluorescence spectra into one. In this step, outliers are identified by means of the shot noise. Furthermore, a fitting routine which model includes Gaussian functions for the fluorescence lines and exponentially modified Gaussian (EMG) functions for the scattering lines (with long tails at lower energies) is proposed to extract the line of interest from the fluorescence spectrum.Additionally, the fitting model has an EMG function for each scattering line (elastic and inelastic) at incident energies where they start to be discerned. At these energies,the data reduction is done per detector column to include the angular dependence of scattering.In the second part of this thesis, an automatic method for texts separation on palimpsests is presented. Scanning x-ray fluorescence is performed on the parchment, where a spectrum per scanned point is collected. Within this method, each spectrum is treated as a vector forming a basis which is to be transformed so that the basis vectors are the spectra of each ink. Principal Component Analysis is employed as an initial guess of the seek basis. This basis is further transformed by means of an optimization routine that maximizes the contrast and minimizes the non-negative entries in the spectra. The method is tested on original and self made palimpsests.
|
The record appears in these collections: |
PDF