A New Method for Segmenting Individual Trees from the Lidar Point Cloud

Abstract

Light Detection and Ranging (lidar) has been widely applied to characterize the 3-dimensional (3D) structure of forests as it can generate 3D point data with high spatial resolution and accuracy. Individual tree segmentations, usually derived from the canopy height model, are used to derive individual tree structural attributes such as tree height, crown diameter, canopy-based height, and others. In this study, we develop a new algorithm to segment individual trees from the small footprint discrete return airborne lidar point cloud. We experimentally applied the new algorithm to segment trees in a mixed conifer forest in the Sierra Nevada Mountains in California. The results were evaluated in terms of recall, precision, and F-score, and show that the algorithm detected 86 percent of the trees ("recall"), 94 percent of the segmented trees were correct ("precision"), and the overall F-score is 0.9. Our results indicate that the proposed algorithm has good potential in segmenting individual trees in mixed conifer stands of similar structure using small footprint, discrete return lidar data.

Deciduous and coniferous tree separation based on the raw intensity of laser points

“…Če se osredotočimo na posamezna drevesa, ki jih bomo obravnavali tudi v naši raziskavi, lahko z laserskimi podatki identificiramo vrhove in določimo višine posameznih dreves. Vrhove posameznih dreves v gozdnem sestoju lahko ločimo med seboj z različnimi algoritmi, ki iščejo lokalne maksimume v surovih podatkih ali na digitalnih modelih krošenj (Reitberger et al, 2008;Kim et al, 2009;Li et al, 2012;Mongus in Žalik, 2015). Če k preučevanju geometrijskih odvisnosti med točkami dodamo še preučevanje intenzitete odbitih laserskih točk, lahko raziskave razširimo na preučevanje zmožnosti določitve posameznih drevesih vrst v laserskih podatkih (Holmgren in Persson, 2004;Moffiet et al, 2005;Kim et al, 2009;Ørka et al, 2009;Korpela et al, 2010).…”

Mapping forests with Lidar provides flexible, accurate data with many uses

“…The ability to delineate individual trees from a Lidar point cloud has been proven for heterogeneous and complex forests such as oak savanna (Chen et al 2006;Chen et al 2007) and mixed-conifer stands (Li et al 2012). Delineating the individual trees is done by segmenting the Lidar-derived canopy height modelthe raster image interpolated from Lidar points depicting the top of the vegetation canopy (e.g., Chen et al 2006) -by delineating the trees directly from the point cloud (Li et al 2012) or by a combination of these methods (Jakubowski, Li et al 2013).…”

“…Delineating the individual trees is done by segmenting the Lidar-derived canopy height modelthe raster image interpolated from Lidar points depicting the top of the vegetation canopy (e.g., Chen et al 2006) -by delineating the trees directly from the point cloud (Li et al 2012) or by a combination of these methods (Jakubowski, Li et al 2013). After accurate segmentation, relationships can be derived between Lidarand field-measured structural attributes such as tree height, crown diameter and canopy base height, which are directly measured, and basal area, diameter at breast height, wood volume, biomass and species type, which are derived by correlations (Chen et al 2006;Chen et al 2007).…”

Individual Tree of Urban Forest Extraction From Very High Density Lidar Data

“…For example, spatial error can be introduced during the interpolation process from the point cloud to the gridded height model (Guo et al, 2010), which can decrease the accuracy of tree segmentations and relevant measurements (Li et al, 2012). To increase the accuracy of detection, Persson et al applied a Gaussian function to elevation model (Persson et al 2002), but the use of smoothing filter cause to estimate the tree height incorrectly (Tiede et al, 2005).…”

“…But it's hard to define this threshold, especially in dense forests. Inappropriate threshold may cause to under-and/or over-segmentation errors (Li et al, 2012). Another problem is that points must participate in the segmentation and classification processes one by one, that in itself rises the computation time.…”