The development of Unmanned Aerial System (UAV) makes it possible to take high resolution images of trees easily. These images could help better manage the orchard. However, more research is necessary to extract useful information from these images. For example, irrigation schedule and yield prediction both rely on accurate measurement of canopy size. In this paper, a workflow is proposed to count trees and measure the canopy size of each individual tree. The performances of three different methods to classify tree canopies are compared. Then morphological methods are used to filter grass patches and separate the trees from each other. Finally, the number of trees and the size of tree canopies are obtained.