Abstract
Crop damage and monetary losses occur due to plant diseases. At specified periods, farmers closely monitor their crops in the field to look for infections or diseases. In place of manual diagnosis, computers have been used to provide computerized tracking and recognition of a variety of diseases.This study makes a contribution to denoising algorithms that improve contrast, edge details and picture details. To isolate the leaf affected area, segmentation was carried out. Image fusion is performed using the segmented image obtained from several segmentation techniques, and features are extracted according to structural, texture, and color criteria. Furthermore, an image fusion technique based on Discrete Shearlet Transforms (DST) is applied to improve imaging quality while reducing redundancy. The color, texture, and structural elements of the fused images are retrieved and fed into the Artificial Neural Network (ANN) for classification, resulting in improved performance. When compared to other classification methods (SVM, MSVM, FFNN, and RFNN), the accuracy required for the DST-based fusion approach and the Radial Basis Function Neural Network is relatively high. The classification accuracy was 99.35%, with 89.24% sensitivity, 94.67% specificity, and 90.12% precision.

This publication has 10 references indexed in Scilit: