Cereal Chem 63:52-56   |  VIEW ARTICLE
Discrimination Between Wheat Classes and Varieties by Image Analysis.

I. Zayas, Y. Pomeranz, and F. S. Lai. Copyright 1986 by the American Association of Cereal Chemists, Inc. 

Image analysis was used on-line to discriminate variables of grain morphology to differentiate among individual kernels of four hard red winter (HRW) and four soft red winter wheat cultivars, and between two HRW and two hard red spring wheat cultivars. Individual kernels and sets of kernels were differentiated using software developed in our laboratories. The main objective was to differentiate rapidly, using a minimum of critical and simple parameters. Differentiation was based on the development of canonical equations for comparing kernels on the basis of six parameters: width, length, length ratio, tangent, sine, and length of arc of parabolic segment. The program was verified by SAS multivariate discriminant analysis using the same variables extracted directly from the image of a kernel. The method was also used for discrimination of mixtures of wheat varieties from the three classes. For mixtures of three varieties of either HRW or of soft red winter wheat, the average percentages of correctly classified kernels were 85% for training set and 83% for experimental set samples. For mixtures of two varieties of either HRW or of hard red spring wheat, the average percentages were 78% for calibration samples and 77% for test samples.