Klasifikace lesních dřevin s využitím hyperspektrálních dat z bezpilotních letadel

Abstract

This thesis deals with the classification of forest tree species using hyperspectral data from unmanned aerial vehicles. The aim is the processing of data from the Resonon Pika L hyperspectral camera mounted on the DJI Matrice 600 Pro hexacopter and the subsequent classification of forest tree species. The thesis is divided into two parts - the first theoretical part explains for example the physical nature, methods of processing hyperspectral data from UAVs (including radiometric, atmospheric and geometric corrections) and selected classification methods. The practical part consists of a description of the camera operation, the creation of flight plans, data acquisition, and above all a detailed description of the processing of the acquired data, classification and subsequent evaluation. Finally, the whole work is summarized, with recommendations and suggestions for improvement.