Airborne lidar has been widely used for forest characterization to facilitate forest ecological and management studies. With the availability of increasingly higher point density, individual tree delineation (ITD) from airborne lidar point clouds has become a popular yet challenging topic, due to the complexity and diversity of forests. One important step of ITD is segmentation, for which various methodologies have been studied. Among them, a long proven image segmentation method, mean shift, has been applied directly onto 3D points, and has shown promising results. However, there are variations among those who implemented the algorithm in terms of the kernel shape, adaptiveness and weighting. This paper provides a detailed assessment of the mean shift algorithm for the segmentation of airborne lidar data, and the effect of crown top detection upon the validation of segmentation results. The results from three different datasets revealed that a crown-shaped kernel consistently generates better results (up to 7 percent) than other variants, whereas weighting and adaptiveness do not warrant improvements.

The study evaluates five existing segmentation algorithms to determine the method most suitable for individual tree detection across a species-diverse forest: raster-based region growing, local maxima centroidal Voronoi tessellation, point-cloud level region growing, marker controlled watershed and continuously adaptive mean shift. Each of the methods has been tested twice over one mixed and five single species plots: with their parameters set as constant and with the parameters calibrated for every plot. Overall, continuous adaptive mean shift performs best across all the plots with average F-score of 0.9 with fine-tuned parameters and 0.802 with parameters held at constant. Raster-based algorithms tend to achieve higher scores in coniferous plots, due to the clearly discernible tops, which significantly aid the detection of local maxima. Their performance is also highly dependent on the moving size window used to detect the local maxima, which ideally should be readjusted for every plot. Crown overlap, suppressed and leaning trees are the most likely sources of error for all the algorithms tested.