Eggplant leaf disease detection and segmentation using adaptively regularized multi Kernel-Based FuzzyC-Means and Optimal PNN classifier

Abstract

Leaf diseases affect both the quantity and quality of crops in agricultural production. Early detection is preventing the plant from diseases. Therefore, in this paper, optimal probabilistic neural network (OPNN) based plant disease classification is proposed. At first, RGB transformation is performed to extract the green band of the eggplant leaf. Then, for the extracted green band; pre-processing is carried using median filter. After pre-processing, the features such as shape, color and vein are extracted. Then the extracted features are fed to the PNN classifier to classify an image as normal or abnormal. To enhance the PNN classifier, the weight values are optimally selected using Binary Crow Search Algorithm (BCSA). Finally, the affected portions are segmented using Adaptively Regularized multi Kernel-Based Fuzzy𝐶-Means (ARMKFCM). This research work is compared with other existing techniques through several performance metrics to show the superiority of our proposed methodology.