Discrimination of some cultivars of durum wheat (Triticum durum Desf.) using image analysis

Abstract:Image analysis technique is application of computer science to get a large amount of information from a digital image. This research was aimed to investigate the potential of morphological features for discrimination of five cultivars of durum wheat (Triticum durum Desf.) including: showa, Altar 84, Altar 84-3, Dipper and Bushen. Algorithms for selecting useful morphological features were developed based on a stepwise discriminate analysis and five feature sets were used for linear discriminate analysis. The results showed that Extent and Aspect Ratio were the most and least important features in discrimination of the cultivar seeds, respectively. Discrimination using the 11 morphological features resulted in the best classification accuracy (67.66%) so that the accuracy of 58.62%, 100, 51.72%, 62.96% and 65% was obtained for Altar 84, Altar 84-3, Dipper, Bushen and showa, respectively. The results emphasized the potential of image analysis in discrimination of durum wheat cultivar seeds.