Return to search

Quantitative near-infrared imaging spectroscopy of fruit

In this work the development and construction of a new near-infrared imaging spectrometer utilising inexpensive components is presented. The instrument is applied to measuring the distribution of constituents in fruit, in particular, soluble solids and dry matter in kiwifruit. The spectrometer, designed for collecting spectral-images of fruit, uses a diffraction grating for spectral dispersion, a scanning slit system for spatial resolution and a CCD detector to allow images of samples to be examined in different wavebands. The system described has a spectral range of 650 nm to 1100 nm at a resolution of better than 5 nm. Each spectral-image contains 150 x 242 pixels on the spatial plane and 755 pixels for the spectral ads. An interchangeable imaging lens enables investigation of samples at different scales. With a 58 mm camera lens objects up to 50 mm x 50 mm may be imaged at a spatial resolution of up to 2 line-pairs per millimetre (0.5 millimetre). One difficulty with calibration using spectral-images is that the independent measurements of concentration often come from areas much larger than a pixel. For spectral-imaging to be useful it is important that the calibration can estimate concentration at the pixel level. In effect this means estimates calculated using the model are more accurate than the independent measurements used in calibration. Using synthetic spectra and an investigation of the principal component regression algorithm, the ability of a calibration, from spectral-images, to estimate these pixel level concentrations has been demonstrated. To calibrate the instrument for measuring soluble solids concentration and dry matter content in kiwifruit, reflectance spectra from 650 nm to 1100 nm were collected from cut sections of 200 fruit. A near-infrared calibration was obtained by recording the soluble solids concentration and dry matter content of plugs extracted from thin slices of the fruit and relating these to the spectra collected. A prediction error of 1.2 °Brix over a range of 4.7 - 14.1 °Brix was achieved. It was not possible to obtain a useful model for estimating dry matter due to interference from specular reflections off free juice in the fruit. The models developed were used to show the spatial distribution of soluble solid concentrations over cut sections of fruit.

Identiferoai:union.ndltd.org:ADTP/247015
Date January 1999
CreatorsMartinsen, Paul
PublisherResearchSpace@Auckland
Source SetsAustraliasian Digital Theses Program
LanguageEnglish
Detected LanguageEnglish
RightsItems in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated., http://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htm, Copyright: The author

Page generated in 0.0081 seconds