A rapid non-destructive Image Analysis (IA) technique was developed for the determination of maize kernel endosperm vitreousness. Kernels were analysed using a Leica Q-Win Q500 IW-DX Image Analyser fitted with Leica Q-Win software and connected to a Sony XC-75 CCD camera. Kernel translucency measurements were optimised by using a light system that involved positioning whole kernels on top of a mask containing round illuminated areas (circles), smaller than the projected areas of the kernels, allowing light to shine through the kernels only. Correction factors allowing for constant illumination of kernels were developed to adjust for kernel size variation in relation to constant light area. Similarly, a correction factor for the effect of kernel thickness on detected translucency values were developed. Significant correlations were found between corrected translucency values and vitreous and opaque endosperm yields as determined by hand dissection. These were: translucency as a percentage of the whole kernel and vitreous endosperm (mass%) (Translucency 1), r = 0.77, p<0.00001, and Translucency 1 and opaque endosperm (mass%), r = -0.72, p<0.00001 for white maize. Similar correlations were found for translucency as a percentage of endosperm (Translucency 2). Correlation coefficients increased significantly after kernel thickness corrections. Significant negative correlations were also found between corrected translucency values and Floating Number. For yellow maize, Translucency 1 correlation coefficients was r = 0.78, p<0.00001 and r = -0.71, p<0.00001 respectively with similar correlations for Translucency 2. Correlations were obtained after applying both correction factors for exposure and thickness. The IA technique was evaluated for predicting the yield of vitreous endosperm products during dry maiz~ milling in laboratory and industrial-scale milling trials. Significant positive correlations were found between corrected translucency values and yields of milling products from vitreous endosperm. Experiments using a laboratory-scale experimental roller milling test without a degerming stage produced the following correlations: between Translucency 1 and semolina yield (mass%), 0.74, p<0.001 and Translucency 2 and semolina yield (mass%), 0.70, p<0.001. For industrial-scale milling, a Bühler industrial-scale maize mill (3 tons per hour) was used. The correlation between Translucency 1 and extraction at degermer (degermer overtail yield) was 0.93, p<0.0001. There was a similar correlation for Translucency 2. Yellow maize was degermed using a pilot-scale Beall-type degermer and the correlation between Translucency 1 and flaking grits> 3.9 mm was 0.67, p< 0.001. The IA technique permits the non-destructive analysis of maize endosperm translucency on large samples of single kernels. It is suitable for rapid quantification of maize endosperm contents and predicting dry maize milling performance, as kernel translucency was significantly correlated with vitreousness in all instances. With further development of specific hardware and software, the technique has potential as an on¬line maize kernel classification system in industrial mills. As the method is non¬destructive, it is also suitable for classification of maize seed breeding material. It is also a potential method for the measurement of maize opacity as used by the wet milling industry, where opacity (the opposite of vitreousness) is related to maize starch yield. / Thesis (PhD(Food Science))--University of Pretoria, 2006. / Food Science / unrestricted
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:up/oai:repository.up.ac.za:2263/24228 |
| Date | 30 April 2005 |
| Creators | Erasmus, Corinda |
| Contributors | Prof J R N Taylor, upetd@up.ac.za |
| Source Sets | South African National ETD Portal |
| Detected Language | English |
| Type | Thesis |
| Rights | © 2003, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. |
Page generated in 0.002 seconds