Abstract

Apple mosaic is a prevalent disease worldwide, causing premature leaf shedding, severe yield reduction, and shortened lifespan of fruit trees. Visible light-near-infrared (VIS-NIR) hyperspectral imaging (HSI) has emerged as a rapid and non-destructive technique for detecting leaf diseases, playing a significant role in plant disease prevention and control. In this study, HSI data was collected on apple leaves infected with the Apple mosaic virus. Gabor images were generated using a three-dimensional Gabor filter. A method based on Simple Linear Iterative Clustering and the Kennard-Stone algorithm was proposed to extract and select representative samples of superpixel scales. Subsequently, an Equilibrium Optimizer was used to select optimal spectral features, and the Fisher Score was utilized to choose optimal Gabor features. Finally, a Support Vector Machine, Multilayer Perceptron, and Random Forest were trained using the optimal features to establish a disease area detection model. The results indicate that the combination features and the Random Forest model achieved the best classification performance, with an overall accuracy of 99.38% and an area under the curve of 0.9985. In conclusion, applying Gabor features effectively improves identification performance, enabling accurate identification of apple mosaic disease areas.