Fungal Disease Detection in Maize Leaves Using Haar Wavelet Features

Agriculture is the backbone of Indian economy. Diseases in crops are causing huge loss to the economy. Only early detection can reduce these losses. Manual detection of the diseases is not feasible. Automated detection of plants diseases using image processing techniques would help farmers in earlier detection and thus prevent huge losses. Maize is an important commercial cereal crop of the world. The aim of this study is the detection of common fungal diseases, common rust, and northern leaf blight in maize leaf. The proposed system aims at early detection and further classification of diseases into common rust, northern leaf blight, multiple diseases, or healthy using first-order histogram features and Haar wavelet features based on GLCM features. Two classifiers, namely, k-NN and SVM are considered for the study. The highest accuracy of 85% is obtained with k-NN for k = 5 and accuracy obtained with SVM-based classification is 88%.