Identification of Egyptian faba beans (Vicia faba l.) cultivars from grain orphological features using discriminant analysis [electronic resource].

Includes references.

The objective of this study was to develop a statistical classifier to identify Egyptian faba beans cultivars using grain morphological features if they are in mixed case. Five Egyptian faba beans cultivars were tested namely: Giza3, Giza461, Misr1, Nobarya1 and Sakha1. The tested faba beans cultivars were grown during agricultural season 2005 and the moisture content ranged from 5.2 % to 6.3 % (w.b). The results showed that the average length of Giza3, Giza461, Misr1, Nobarya1 and Sakha1 was 16.71, 16.12,15.61, 18.87, and 16.25 mm, respectively. In spit of this converge in length values among cultivars, the efficiency of the develop classifier had reasonable accuracy. The best prediction accuracies was 67.9% when using testing data set with quadratic discriminant function. The corresponding category wise prediction accuracies were 45.5, 62.5, 87.5, 80.0 and 68.8% for Giza3, Giza461, Misr1, Nobarya1 and Sakha1, respectively with test data set using quadratic discriminant function. The development discriminant analysis showed low accuracy with its performance. So, it is recommended to improve the discrimination performance by adding another features to the grain morphological features like features extracted form texture and/or color of faba beans cultivars images obtained by machine vision technique. Whereas, the development discriminant analysis is conceptually simpler, easier to code into computer applications for faba beans (Vicia Faba L.) discrimination purposes in Egyptian grading, packing and cleaning seeds stations.

Summary in Arabic.

1