Characterization of Potato Leaf Disease by Digital Image Processing Technique

Abstract

Potato is a significant staple crop in Bangladesh. The productivity of potatoes decreases by factors such as disease, insect infestation, and rapid variations in climate conditions. The classification of potato leaf disease shows a vital role in preventing a damage of product. To identify the signs of disease immediately appearing in plant, it is essential to use automated detection techniques. If these epidemics are identified at the initial stage and proper activity is selected, the farmers would not suffer from significant financial losses. In this study, the classification of diseases of potato leaf was proposed using a digital image processing technique. The steps followed in this technique were acquisition, pre-processing of image, segmentation of image, feature extraction from image, and disease classification. For image acquisition, the early blight, healthy leaf, and late blight of potato leaf were clicked using DSLR camera. Enhancing the contrast and removing noise, RGB images were pre-processed. The diseased portion, normal portion, and background area was segmented through k-means clustering. Then the diseased portion was converted into a grayscale image. Feature extraction was done using an algorithm known as Gray Level Co-occurrence Matrix(GLCM). The classification of disease was done using Support Vector Machine (SVM). The proposed method achieved in identifying the early blight infected leaves, late blight infected leaves, and healthy leaves, was 95%, 76.5%, and 90%, respectively. Therefore, the proposed potato leaf disease finding by means of image processing may be a successful technique nowadays.