![]() | Only 14 pages are availabe for public view |
Abstract Recent technology trends to miniaturize spectrometers have opened the doors for mass production of spectrometers and for new applications that were not possible before and where the spectrometer can possibly be used as a ubiquitous spectral sensor. However, with the miniaturization from large reliable bench-top instrument to chip-size miniatur- ized spectrometers leads to new issues and challenges which are introduced into building multivariate spectroscopic models based on these spectrometers. The purpose of this thesis is to study the feasibility of building multivariate models based on miniaturized Fourier Transform Near-Infrared (FT-NIR) spectrometers, and determine the issues emerged and propose appropriate solutions for them. The thesis presents some classification models with di↵erent natures, and each model introduces a new challenge associated with our proposed handling, the models are textiles type classification, co↵ee classification according to ca↵eine level, species type classification and milk classification according to fat level models. In addition, the thesis presents two regression applications with commercial standards, namely, milk regression application and health care application. In this work, we proposed a calibration transfer technique to mitigate the e↵ect of unit- to-unit variations, variations due to changing the measurement setup and variations due to changing the measurement medium. The technique shows a significant improvement in the performance of the models subjected to this kind of variations, like in the milk classification model and the regression models of the health care application. |