Assessment of the Scots pine (Pinus sylvestris L.) crowns density based on multispectral images obtained by unmanned aerial vehicle

This paper presents the use of non-metric multispectral digital images analysis acquired by cameras carried by unmanned flying vehicles (UAVs) to assess the density of crowns in Scots pine stands. Images were acquired in October 2011. During the field data acquisition, 272 pine trees were inventoried and classified to 10 classes based on the crown density. These results were compared with the average pixel brightness in four spectral channels B, G, R and IR, collected in three variants – the whole crown, dense part of the crown (the so-called rejection of “outlier” branches) and the central part of the crown with 0.63 m radius. Studies have shown that brightness of images of tree crowns belonging to different classes of density varies insignificantly. This especially concerns the class of trees with a very high degree of defoliation (70–100%) and dead trees. The degree of stabilization can be observed in class 5, which means that a further increase in crown density does not increase the spectral reflectance. Analyzes revealed strong similarity between crowns variants: spectral images obtained for the whole or a part of the crown do not differ from each other. Ultimately, it is difficult to determine which variant is the best for the further classification procedure