Plant Leaf Disease Classification Using Modified SVM With Post Processing Techniques
R. Thyagaraj, T. Y. Satheesha, Sathish Bhairannawar
Abstract
In order to prevent yield loss and quality of the agricultural product, early detection of diseases in plant is essential. Using visually apparent patterns on the plant, numerous studies on plant illnesses have been extensively investigated to discover abnormalities in plant growth. A sustainable agriculture depends on plant monitoring and disease detection. However, because they demand precise and real-time diagnosis, monitoring plant diseases manually is highly challenging. Plant disease diagnosis frequently uses image processing, which includes picture acquisition, preprocessing, segmentation, feature extraction, and classification. In this paper, an efficient plant disease classification using SVM Classifier is proposed. Firstly, the adaptive histogram technique is applied to the image and the required part is segmented using Otsu’s segmentation. The segmented output is applied to SVM to classify between the normal and diseased images. Further, the soft thresholding is applied to improve the accuracy. The proposed method is tested for standard database to obtain 95% accuracy compared to existing techniques.