ANN-BASED CLASSIFIER TRAINED BY BAYESIAN REGULARIZATION FOR WHEAT GRAINS THROUGH COLOUR FEATURE

Authors : Berat YILDIZ, Abdurrahim TOKTAŞ, Enes YİĞİT, Ahmet KAYABAŞI, Kadir SABANCI, Mustafa TEKBAŞ

Pages : 46-53

View : 32 | Download : 9

Publication Date : 2017-12-27

Article Type : Research Paper

Abstract :In this paper, a color feature-based classification of the wheat grains into bread and durum using artificial neural network insert ignore into journalissuearticles values(ANN); model with bayesian regularization insert ignore into journalissuearticles values(BR); learning algorithm is presented. Images of 200 wheat grains are taken by a high resolution camera in order to generate the data set for training and testing processes of the ANN-BR model. Data of 3 main colour features insert ignore into journalissuearticles values(R, G and B); for 200 wheat grains insert ignore into journalissuearticles values(100 for durum and 100 for bread); are acquired for each grain using image processing techniques insert ignore into journalissuearticles values(IPTs);. Features of R, G and B are separately determined by taking arithmetic average of the pixels within each grain. Several colour features of R/TRGB, G/TRGB, B/TRGB, R-G, G-B and R-B where TRGB is the total of R+G+B are reproduced. Then ANN-BR model input with the 9 colour parameters are trained through 180 wheat grain data and their accuracies are tested via 20 data. The ANN-BR model numerically calculate the outputs with mean absolute error insert ignore into journalissuearticles values(MAE); of 0.0060 and classify the grains with accuracy of 100% for the testing process. These results show that the ANN-BR model can be successfully applied to classification of wheat grains.
Keywords : Classification, wheat grains, image processing technique, artificial neural network, bayesian regularization learning algorithm