seedcount presentation 002

Screenings and Roundness: 

Screenings and Roundness DETERMINING SCREENING FRACTIONS AND KERNEL ROUNDNESS WITH DIGITAL IMAGE ANALYSIS Presented by Bruce Armstrong University of Ballarat

Introduction: 

Thickness measurement an issue with digital image analysis (DIA) Introduction Basic Concepts: A Screening Assortment is made by separating the kernels into various groups by their thickness. The kernels are sorted by shaking them through a stack of screens with slots of decreasing widths. (eg 2.8, 2.5, 2.2 and 2.0 mm)

Screening: 

Thickness measurement an issue with digital image analysis (DIA) Screening The shaking action allows the kernels to strike the slots “on-edge” (bottom right) and slide through slots that are too narrow for them to pass through when laying on the flat.

Why do Screenings Matter?: 

Grain with too many undersized kernels (thins) will be downgraded to Feed Grade. Slotted screens are used to remove thins by most processors. Grain behavior (flour/modification) in a mill or malt house is affected by the kernel size and thickness. Large-kernelled grain looks more appealing to many purchasers. Large kernels are usually expected to yield more flour/extract. Why do Screenings Matter?

What is Roundness?: 

It is the three-dimensional sphericity of the kernels This work uses the following novel concept of Roundness: R = (W/L + T/L + T/W)/3 A basketball has a roundness of 1, while a long, thin grass-seed has a roundness of 0.25 W = Width, L = Length, T = Thickness What is Roundness?

Does Roundness Matter?: 

Grain performance can be affected by kernel shape as well as size. Round kernels should have a higher flour/extract yield than elongated kernels. But it is extremely tedious to determine roundness by classical methods (calipers) Does Roundness Matter?

Research Questions: 

Can Digital Image Analysis (DIA) be used to directly measure kernel thickness? Can DIA be used to estimate Kernel Mass, Roundness and Screening Assortments? Can DIA Roundness and Screenings be used to predict wheat flour yield and malted barley soluble extract? Research Questions

Kernel Morphology: 

Kernel Morphology Kernel structure is important to prediction of yields. Crewe and Jones (1951) demonstrated that cereal bran thickness is stable over a wide range of kernel sizes. The following simplified table assumes that the kernel shape is ellipsoid and the bran/husk layer coating the barley is uniformly 0.08 mm thick.

Theoretical Effects of Kernel Shape and Size: 

Theoretical Effects of Kernel Shape and Size

Materials and Methods: 

Individual kernel data from cultivars grown in Victoria. Wheat flour data from 42 cultivars grown in New South Wales and Queensland Barley extract data from 41 cultivars grown in Victoria, South Australia and Western Australia Standard IOB, Allied Mills and Joe White Malting methods used for non-DIA testing. DIA work performed on a SeedCount system. Materials and Methods

SeedCount Screenshot: 

SeedCount Screenshot Demonstrations available at the Graintec booth

Kernel Shape and DIA: 

Screening separates seeds by thickness. Thickness is rarely seen when seeds are spread on a flat surface. Thickness is needed to assess roundness. Kernel thickness is also needed for calculation of “virtual kernel” mass. Special “flat-edge” tray is needed to hold some seeds “on-edge” to allow the direct measurement of thickness. Kernel Shape and DIA

Barley Flat-edge Tray Detail: 

Wide “flat” Section Barley Flat-edge Tray Detail Narrow “on-edge” Section

Virtual Kernels: 

Virtual Kernels Required for the calculation of kernel roundness and mass. Created by matching the smallest kernel in the narrow “on-edge” section with the smallest kernel in the wide “flat” section. DIA thickness is used to allocate each virtual kernel (and its mass) to a screening group. Allows the percent mass in each group to be assessed.

Thickness calc: 

Thickness calc

Thickness cultivar: 

Thickness cultivar

Mass Estimates: 

“Virtual” three-dimensional kernels used. Multivariate, multigroup approach Minimizes errors from individual measurements Mass Estimates

Mass calc: 

Mass calc

Mass est by cultivar: 

Higher res and larger sample sizes would further improve accuracy Mass est by cultivar

Screenings Assortment: 

Screenings Assortment Previous data on a seed by seed comparison Following results are for full tray “bulk” samples of 600 to 1000 kernels. Samples run in SeedCount and then screened with certified slotted screens. Std errors are for 95% confidence intervals. Many cultivars in the validation tests are different to the developmental kernel samples.

Screenings graph: 

Screenings graph

Barley Screen Assortment Accuracy: 

Barley Screen Assortment Accuracy

Wheat Screen Assortment Accuracy: 

Wheat Screen Assortment Accuracy

Roundness - Calipers vs DIA: 

Correlation (r) on a seed by seed basis: Barley = 0.91, std error 0.02 Wheat = 0.81 std error 0.03 Cultivar avg basis: Wheat R2 = 0.90, SEE 0.01 Avg Roundness of barley cultivars is 0.43 to 0.53. Avg Roundness of wheat cultivars is 0.56 to 0.64. Roundness - Calipers vs DIA

Screenings, Roundness and Yield: 

Screenings, Roundness and Yield

Flour Yield Prediction: 

Estimated Dirty Wheat flour Extraction = - 2.819 + 0.2092* HW + 1.793 * Roundness – 0.4009 * Screening + 0.6252 * CW + 0.1044 * Hardness – 0.592 * Screenings OT HW = Hectoliter Weight (clean mini SeedCount) Roundness -DIA Screening = DIA 2.0 to 2.2 mm group CW = Chondrometer Weight (dirty 500 ml sample – usually DIA) Hardness = resistance of kernel to crushing – non DIA Screenings OT = Overtail, large pieces of dockage material. (normally by DIA system) Flour Yield Prediction

Yield prediction graph: 

Yield prediction graph

Conclusions: 

DIA systems using “flat-edge” trays can measure kernel thickness. DIA systems using “flat-edge” trays can estimate screening assortments and kernel roundness. DIA Roundness and Screenings can be used to predict wheat flour yield but not soluble barley extract. Conclusions

Role of DIA in Quality Assessments: 

Has the potential to make objective, repeatable, storable visual tests Kernel Weight and Counting are already highly accurate (Armstrong et al, 2001) Fast dimensional, area and roundness evaluations Can estimate Screening assortments More complex colorimetric assessments such as blackpoint, fusarium possible. Role of DIA in Quality Assessments

Acknowledgements: 

This Research was supported by: Australian Research Council and the Wrightson Research Co-authors : Peter Aldred, John Dines, Jarrod Gooden, Rob Greig and Marvin Weiss Acknowledgements

Copyright Details: 

Copyright  2003 All rights reserved. This presentation was made at the 2003 Barley Technical Symposium/Royal Australian Chemical Institute-Cereal Chemistry Division Conference held during September at Adelaide, Australia and the paper has been published in both the BTS and CCD Conference Proceedings. www.seedcount.com.au Copyright Details