A new method for detecting individual trees in aerial LiDAR point clouds using absolute height maxima

Data acquired from aerial laser scanner systems are increasingly used for detecting individual trees in operational inventories. In conventional analyses, tree detection is often performed on raster models that use local height maxima filters; an option that is likely to accumulate important errors. In order to reduce errors and improve the detection of individual trees, a new method is proposed that uses an Absolute Height Maxima (AHM) filter applied on the original point clouds obtained from Aerial Laser Scanning (ALS). ALS point clouds at a density of 2 to 4 points per square meter were acquired over forest stands in Hyrcanian forests. In the new method, false trees and commission errors were automatically found and excluded. To evaluate the efficiency of this new method, 121 sample trees in the field were located, with a DGPS and a mapping camera. The height and crown radius of the sample trees were also measured. The field-surveyed variables were compared to the closest detected tree, with an overall detection accuracy of 75.2%. The initial results of this analysis allowed us to hypothesize that a higher detection of tree may be expected with larger densities.