Study of algorithms for analysis of xrf spectra to automate inspection of carpets
MetadataShow full item record
The objective of this thesis is to categorize carpet types according to their XRF spectra and verify if further classification of carpets is possible for use of an XRF analysis system in the carpet manufacturing line. This thesis consists of (1) implementing and studying effective algorithms for automated analysis of X-ray spectra, (2) comparing known algorithms for X-ray spectra analysis, and (3) implementing our own algorithm for classification of carpets spectra obtained for further fluorine online analysis of XRF inspected carpets. This research is intended for quick and accurate automated analysis of raw XRF spectra and matching analysis results to a database of XRF spectra of raw carpets. The research uses spectrum signal processing and spectrum analysis regarding efficacy of combined methods for XRF inspected carpets. X-Ray Fluorescence is a key technology for detection of chemical elements. Fluorine is a key element for carpet's quality. XRF has been chosen to be a potential candidate to measure fluorine since it is a versatile tool for low concentration element detection. Due to specific XRF background spectrum for each different carpet type, carpet samples may need specific calibrations for further computation of carpet fluorine concentration. Automating the detection of the carpet type is intended to help in automating the XRF calibration.