Identification of Wheat and Foreign Matter Using Artificial Neural Network and Genetic Algorithm

Abstract

AbstractOne of the major issues with wheat grains is the presence of foreign matter in the same, which significantly reduces the quality and marketability of the product. Traditional methods of detecting foreign matter in wheat grains using manual inspection are always associated with some typical problems such as time-consumption and inconsistency. In order to circumvent these problems, machine vision has been considered as one of the very effective alternative for such applications. The main task in the present work is to detect the foreign matter in wheat grains. This study explores the potential of identifying extraneous material in wheat kernels using hybrid approach involving the use of Artificial Neural Network (ANN) and Genetic Algorithm (GA). Images of wheat kernels and foreign matter are preprocessed and sixty one features have been extracted. The extracted features were classified into two classes. The first aims at identifying wheat grains and the second one, identifies foreign matter in wheat grains. Best classification accuracy has been achieved with hybrid GA-ANN model. The highest accuracy achieved with the proposed hybrid model is found to be 98.7% for testing phase. Accordingly, the proposed system in this work accurately detects foreign matter in wheat kernels. The results of present study are quite promising. KeywordsArtificial Neural NetworkGenetic algorithmMachine visionMorphologicalColorTexture and image processing