With the increasing population and the growing demands for rice, wheat, maize and other food products, the production should be increased in quantity and quality as well. India has a high production of wheat both for the local consumption and towards export to other countries as well. More than 100 varieties of wheat are bred; the varietal purity of wheat grain is a challenge to be handled. The objective of this study is to form an automatic wheat variety identification algorithm using primary morphological features. 280 images of wheat samples were acquired and analyzed. The specific characteristics required for discrimination of varieties, such as RGB value, area, centroid, entropy and histogram of oriented gradient features were chosen and a classification algorithm using Artificial neural network was developed. 60% of the images were used for training, and 40% were used to evaluate the performance of the model. A classification accuracy of 91.86% was obtained for classification of 28 wheat varieties.
Volume 11 | 01-Special Issue
Pages: 82-86