Comparison of statistical and neural network methods for classifying cereal grains using machine vision
1999
Luo, X. | Jayas, D.S. | Symons, S.J.
Two statistical and one neural network classifiers were applied and empirically compared for the classification cereal grain kernels (e.g., Canadian Western Red Spring (CWRS) wheat, Canadian Western Amber Durum (CWM)) wheat, barley, rye, and oats) and for the classification of healthy and six types of damaged (e.g. broken, grass-green/green-frosted, black-point/smudge, mildewed, heated, and bin/fire-burnt) CWRS wheat kernels, using selected morphological and color features extracted from the grain sample images. For the classification of cereal grain kernels and the classification of healthy and damaged CWRS wheat kernels, the k-nearest neighbor statistical classifier and the multilayer neural network (MNN) classifier gave similar and the best classification results. Using a k-nearest neighbor classifier with a selected set of 15 morphological and 13 color features, the average classification accuracies were 98.2, 96.9, 99.0, 98.2, and 99.0% for CWRS wheat, CWAD wheat, barley, rye, and oats, respectively, when trained and tested with three different training and testing data sets. Using a k-nearest neighbor classifier with a selected set of 24 color and four morphological features, the average classification accuracies were 92.5 (healthy), 90.3 (broken), 98.6 (mildewed), 99.0 (grass- green/green-frosted), 99.1 (black-point/smudged), 97.5 (heated), and 100.0% (bin/fire-burnt), respectively, when trained and tested with three different training and testing data sets. The classification accuracies achieved using a parametric classifier were lower than the classification accuracies achieved using both the k-nearest neighbor and the MNN classifiers.
显示更多 [+] 显示较少 [-]