Remote sensing by drones of areas infected by Xylella Fastidiosa by using multispectral techniques

This paper deals with an accurate and efficient procedure for the detection of trees affected by Xylella Fastidiosa using Unmanned Aerial Vehicles (UAVs) and multispectral techniques. As well known, UAVs can acquire and collect many leaf images to detect the presence of possible disease symptoms. A suitable processing system was developed and implemented to carry out and compare results, and finally to determine the accuracy in disease monitoring. Particularly, a new segmentation algorithm to recognize trees is applied, and images are then classified by using linear discriminant analysis. If well used, the proposed procedure improves both the feasibility of a correct tree individuation, and its sensitivity in detecting infected trees.