Texton Features and Deep Belief Network for Leaf Disease Classification
M Anandkumar
Abstract
A world without agriculture would be very difficult for all living things because agriculture is an important part of our daily life. Day by day, agriculture is destroyed due to water scarcity, flood, pollutions, soil erosion, deforestation, plant or leaf disease, and so on. The main impact of agriculture is leaf disease or plant disease, which is caused by some kinds of fungus, bacteria, and viruses. If the farmers do not take proper care, then it causes more problems and reduces the quantity and quality of the product. This is the main reason that the detection of leaf disease is very important. In this paper, the tomato leaf disease classification is performed by the Deep Belief Network (DBN). The DBN is mainly focused to classify leaf diseases and improves the quality and production quantity of tomatoes. This method filters the unwanted signal from an input image by Gaussian filter, and then, the Region of Interest (ROI) reduces the pixels of the image. After that, the feature extraction extracts the received data, and finally, the DBN classification system classifies the leaf disease. The implementation and analysis of the proposed method exposed that the DBN classifier for tomato leaf disease classification yields a better performance of 98.97%.