Skip to main content
ARS Home » Plains Area » Manhattan, Kansas » Center for Grain and Animal Health Research » Grain Quality and Structure Research » Research » Publications at this Location » Publication #99931
Title: CAN DIGITAL IMAGE ANALYSIS BE DEVELOPED INTO A REFERENCE METHOD FOR DETERMINING STARCH SIZE DISTRIBUTION

Author
item Bechtel, Donald
item Martin, Charles
item Wilson, Jeff

Submitted to: Cereal Foods World
Publication Type: Abstract Only
Publication Acceptance Date: 6/24/1999
Publication Date: N/A
Citation: Bechtel, D.B., Martin, C.R., Wilson, J.D. 1999. Can digital image analysis be developed into a reference method for determining starch size distribution. Abstract No. 338 in: 1999 AACC Annual Meeting Program Book. p.293. Meeting Abstract.

Interpretive Summary: To be presented at the 84th AACC Meeting held October 31-November 3, 1999, in Seattle, WA.

Technical Abstract: Light microscopy has long been the method of choice for measuring microscopic particles. The process is slow, labor intensive and limited numbers of particles can be analyzed, however. Automated image analysis systems have been developed which addressed some of the concerns associated with these earlier methods. Other important issues have yet to be addressed, in particular, how particles that touch the edge of field of view are handled. Previously, image analysis systems have either counted and measured the partial views of these particles or eliminated them from the analysis. Problems in analyzing starch granules are complex because of the wide range of sizes, from less than one micrometer to more than 30 micrometers in diameter. The larger the particle or higher the magnification used the more likely that a particle will be touching the edge of the field of view. We found that the magnitude in which large type A starch granules can touch the edge of field of view can approach 50 %. Although type A granules generally occur at less than a 7 % frequency, their large size contributes most of the total starch mass. Even small errors associated with counting the number of type A granules, therefore, can greatly influence the total mass attributed to them. Using log normal distributions of the data we have developed mathematical approaches to correct the errors associated with particles touching the edge phenomena. Image analysis can be used as a reference method for starch size distribution determinations.