Rapid identification of barley varieties by machine vision examination of seed

Abstract

The technique of image analysis of grain characteristics has some potential for distinguishing between varieties of wheat. In this project, a dedicated, prototype wheat grain image analyser, developed by NIAB at the Official Seed Testing Station (OSTS) for England and Wales, with funding from Twyfords seeds, was used to obtain binary /images of seed samples taken from five varieties of two-rowed winter barley.

Although there were consistent differences between Halycon, Igri, Pipkin, Maris Otter and Panda, it was not possible to distinguish the varieties using image analysis. Of the varieties tested, Halycon and Maris Otter cannot be discriminated by means of electrophoresis and Pipkin and Maris Otter cannot be discriminated visually.

Since the project was completed, there have been significant advances in both computing and image analysis. As the varieties used in the project could be distinguished on the basis of some of the 69 quantitative measurements (descriptors) taken for each grain, it is possible that image analysis could now (January 1996) be a viable technique for distinguishing barley varieties.