Wheat Fusarium Head Blight Detection Using UAV-Based Spectral and Texture Features in Optimal Window Size

By combining the spectral and texture features of images captured by unmanned aerial vehicles (UAVs), accurate timely detection wheat Fusarium head blight (FHB) can be realized. This study presents a methodology to select optimal window size gray-level co-occurrence matrix (GLCM) extract from UAV for FHB detection. Host conditions disease distribution were combined construct model, its overall accuracy, sensitivity, generalization ability evaluated. First, sensitive bands UAV-derived hyperspectral obtained, then selected. Subsequently, extracted windows different sizes input classify area severe FHB. According model comparison, was obtained. With collinearity between eliminated, best performance logistic reached, with an F1 score, under receiver operating characteristic curve 0.90, 0.79, respectively, when GLCM 5 × pixels on May 3, 0.83, 0.82, 17 8. The results showed that selection appropriate feature extraction enabled more