Classification of cereal grains using machine vision: II. Color models

A digital image analysis (DIA) algorithm was developed based on color features to classify individual kernels of Canada Western Red Spring (CWRS) wheat, Canada Western Amber Durum (CWAD) wheat, barley oats, and rye. Eighteen color features (mean, variance, and range of red, green, and blue, and hue, saturation, and intensity) were used for the discriminant analysis. Grains from 15 growing regions (300 kernels per growing region) were used as the training data set and another five growing regions were used as the test data set. When the first 10 most significant color features were used in the color model and tested on an independent data set (the test data set where total number of kernels used was 10,500; for CWRS wheat, 300 kernels each were selected for three grades), the classification accuracies of CWRS wheat, CWAD wheat, barley, oats, and rye were 94.1, 92.3, 93.1, 95.2, and 92.5%, respectively. When the model was tested on the training data set (total number of kernels used was 31,500), the classification accuracies were 95.7, 94.4, 94.2, 97.6, and 92.5%, respectively for CWRS wheat, CWAD wheat, barley, oats, and rye.