Using Multispectral Imagery from UAV to Derive Selected Forest Inventory Parameters

Abstract

The following article deals with the usage of Unmanned Aerial Vehicles in the forest inventory. The main goal is to obtain general information about certain forest areas based on deriving data from UAV captured multispectral images. Parameters like the number of individual trees plus their species classification and health condition evaluation are obtained via the usage of deep learning techniques. Additional parameters, like crown diameter, are calculated too based on the segmentation of the digital elevation model. Used techniques are described, evaluated and further development proposed.