Applied 3d Texture Features in Als-based Tree Species Segmentation

Airborne laser scanning (ALS) data are usually considered not to be very informative with respect to tree species, and the information needed regarding these is often obtained by combining the ALS data with spectral image material. This study employed tree species segmentation based on variables derived solely from ALS point clouds to test the ability of height, density, and applied 3D texture variables to describe species-specific characteristics of crown shape and structure. Linear discriminant analysis was used to find optimal combinations of variables within different predictors and segmentation techniques. Segmentation procedures based on different feature sets were compared. Alpha shape metrics, developed for describing tree crowns in ALS data, proved able to discriminate spatially between all three species evaluated and improved final segmentation result. * Corresponding author. This is useful to know for communication with the appropriate person in cases with more than one author.