The influence of visible-shortwave NIR spectroscopy hardware components on prediction of fruit attributes

Overall performance of a visible-shortwave near infrared (NIR) spectroscopy system for a given application will be influenced by component performance, including optical geometry, lamp type and output stability in terms of aging and environmental conditions, detector wavelength range, signal to noise and stability to environmental conditions, model transfer across instruments and model robustness in terms of function in the face of varying temperature and ambient light levels. This thesis investigated the characteristics of individual components in context of their effect on performance of partial least squares (PLS) models of intact fruit total soluble solid content.