DIRT LEVEL ESTIMATION IN PREPARED CANE USING VIS/NIR SPECTROSCOPY

THIS study investigated the use of visual to very near infra-red (VIS/VNIR 400–1100 nm) spectra to determine dirt levels of prepared cane in a laboratory setting. VIS/VNIR spectra contained reflection/absorption features that were just as sensitive to dirt concentration as features in near-infra-red (NIR 1100–2500 nm) spectra. VIS/VNIR spectra could be used to make similarly accurate dirt concentration estimates as NIR spectra. Beer’s model estimates of dirt level were compared with estimates made using multivariate linear regression, both with and without knowledge of dirt type. The best estimates were obtained using multivariate linear regressions with knowledge of dirt type. Dirt type could be identified using discriminant analysis and dirt level estimates made using predicted dirt type were significantly more accurate than ‘global’ estimates. They were also unbiased in their error distributions.