classification of iranian wheat flour by ft-mir spectroscopy based on max-relevance min-redundancy wavelength selection coupled with svm

Different varieties of wheat as one of the strategic crops are cultivated in iran based on the specific geographical and climatic conditions of each area. classification of wheat varieties is important in order to guarantee the final products acquired from wheat flour. fourier transform-mid infrared (ft-mir) spectroscopy as a nondestructive approach combined with chemometrics was employed to classify four varieties of iranian wheat. 160 samples were analyzed and various preprocessing algorithms were used to correct unwanted information. then, principal component analysis (pca) as unsupervised and support vector machine (svm) as supervised models with max-relevance min-redundancy (mrmr) feature selection algorithm were applied to investigate the classification of these varieties. the best result of svm model without feature selection was with s-g+d2+msc preprocessing with 99.4% of accuracy. the output of 100% with svm model and mrmr feature selection algorithm confirmed the capability of ft-mir spectroscopy method for classification of iranian wheat flour varieties.

classification of iranian wheat flour by ft-mir spectroscopy based on max-relevance min-redundancy wavelength selection coupled with svm

different varieties of wheat as one of the strategic crops are cultivated in iran based on the specific geographical and climatic conditions of each area. classification of wheat varieties is important in order to guarantee the final products acquired from wheat flour. fourier transform-mid infrared (ft-mir) spectroscopy as a nondestructive approach combined with chemometrics was employed to classify four varieties of iranian wheat. 160 samples were analyzed and various preprocessing algorithms were used to correct unwanted information. then, principal component analysis (pca) as unsupervised and support vector machine (svm) as supervised models with max-relevance min-redundancy (mrmr) feature selection algorithm were applied to investigate the classification of these varieties. the best result of svm model without feature selection was with s-g+d2+msc preprocessing with 99.4% of accuracy. the output of 100% with svm model and mrmr feature selection algorithm confirmed the capability of ft-mir spectroscopy method for classification of iranian wheat flour varieties.