seedcount presentation 003

Digital Imaging: 

Digital Imaging Using Digital Image Analysis to Accurately Determine the Thousand Kernel Weight of Randomly Distributed Barley, Malt and Wheat. Presented by Bruce Armstrong University of Ballarat

Introduction: 

Why use Thousand Kernel Weights? What is a Thousand Kernel Weight? The mass, in grams, of a thousand kernels of grain. Large, heavy grains tend to have a higher flour yield or more malt extract than small grains. Plump grain is attractive to customers. Important breeding characteristic for cultivar selection. Useful in determining the correct sowing rate for optimum plant densities. Introduction Provides a quantitative measure of the mass of an average kernel. _________________________

Which shipment would you buy?: 

Which shipment would you buy? Which shipment would you buy? TKW = 41.4 TKW = 18.0 TKW = 50.4 TKW = 23.2

Plant Material Analyzed: 

Plant Material Analyzed Wheat: Brennan, Gordon, Kellalac, Meering and Silver Star plus Kellalac screenings >2.8 mm and <2.0 mm Barley: Franklin, Gairdner and Parwan plus Gairdner screenings >2.8 mm and <2.0 mm Malt: Franklin, Gairdner, IS9 and P52 ____________________________

Methods of Counting TKW: 

Hand Counting Hand Counting using Seed Trays Counting with Electromechanical Counters Counting with highly specialized Image Analysis Systems Counting with readily available digital image manipulation equipment. Methods of Counting TKW _________________________

Using Digital Image Processingto determine TKW: 

Touching kernels present counting difficulties. Can be resolved by: Ensuring that all kernels are physically separated Using digital image manipulation to alter scan to separate touching kernels Ignoring all touching kernels Developing a non-destructive algorithm to count separate and touching kernels Using Digital Image Processing to determine TKW _________________________

SeedCount Algorithm: 

SeedCount Algorithm Initially the number and cross-sectional area of the single grains are determined. The average area of the single grains is calculated. This value is used to determine the number of grains in the remaining grain clusters. _________________________

Typical scan of Franklin barley: 

Typical scan of Franklin barley Typical scan of Franklin barley

Eroded Franklin scan: 

Eroded Franklin scan Eroded Franklin scan ready for Scion Image to Count

Eroded Franklin Full: 

Eroded Franklin Full

SeedCount Screenshot: 

SeedCount Screenshot SeedCount Screenshot of Counted Franklin Scan.

SeedCount Full: 

SeedCount Full

Wheat Hand&Tray Graph: 

Wheat Hand&Tray Graph

Wheat Numigral Graph: 

Wheat Numigral Graph

Wheat Scion Image Graph: 

Wheat Scion Image Graph

Wheat SeedCount Graph: 

Wheat SeedCount Graph

Barley Numigral Graph: 

Barley Numigral Graph

Barley SeedCount Graph: 

Barley SeedCount Graph

TKW Method Comparisons: 

TKW Method Comparisons

Scanner Comparisons: 

Scanner Comparisons

Seed Presentation Table: 

Seed Presentation Table

Extra Calcs: 

Extra Calcs As well as calculating TKW (Dry and as-is), Seedcount also calculates screening equivalents, seed area, dockage and the hectoliter weight.

Data Storage: 

Data Storage Data can be stored in a file and imported into Excel. Scan images can also be stored for future reference.

Further Developments: 

Kernel Plumpness (Aspect Ratio) Improve match of screenings equivalents with mechanical screenings Directly determine weather damage, foreign matter and broken and diseased grains. Add calibrations for other seeds, such as rice, maize, soybeans and canola Further Developments _________________________

Acknowledgements: 

Australian Research Council and Wrightson Research for funding. Jarrod Gooden and Alex Severino of Joe White Maltings Samantha Openshaw of Agrifood Technology Paul Panozzo of Goodman Fielder Mills Joe Panozzo of the Victorian Institute of Dryland Agriculture Acknowledgements _________________________

Copyright Details: 

Copyright  2001 All rights reserved. This presentation was made at the 2001 Australian Cereal Chemistry Conference held during September at Sydney, NSW and the paper has been published in the Cereals 2001 Conference Proceedings. www.seedcount.com.au Copyright Details _________________________