Individual tree segmentation with several possible algorithms (see details). The function
attributes a number to each point of the point cloud identifying the detected tree that
the point comes from (in a new treeID column). By default the classification is
done at the point cloud level and the functions retun nothing. However, with some algorithms
it is possible to also return a raster image of the classification. There are currently
4 algorithms implemented. See relevant sections.
lastrees(las, algorithm, ...)lastrees_li(las, dt1 = 1.5, dt2 = 2, Zu = 15, hmin = 2, R = 10)
lastrees_watershed(las, chm, th_tree = 2, tol = 1, ext = 1,
extra = FALSE)
lastrees_dalponte(las, chm, treetops, th_tree = 2, th_seed = 0.45,
th_cr = 0.55, max_cr = 10, extra = FALSE)
lastrees_silva(las, chm, treetops, max_cr_factor = 0.6, exclusion = 0.3,
extra = FALSE)
An object of the class LAS. If missing extra is turned to TRUE
automatically.
character. The name of an algorithm. Can be "dalponte2016",
"watershed","li2012" or "silva2016" (see sections relevant to each
algorithm).
parameters for the algorithms. These depend on the algorithm used (see details about the algorithms)
numeric. Threshold number 1. See reference page 79 in Li et al. (2012). Default 1.5.
numeric. Threshold number 2. See reference page 79 in Li et al. (2012). Default 2.
numeric. If point elvation is greater than Zu, dt2 is used otherwise dt1 is used.
See reference page 79 in Li et al. (2012).
numeric. Minimum height of a detected tree. Default 2.
numeric. Maximum radius of a crown. Any value greater than a crown is good because this parameter does not affect the result. However, it greatly affects the computation speed. The lower the value, the faster the method. Default is 10.
RasterLayer. Image of the canopy. You can compute it with grid_canopy or grid_tincanopy or read it from an external file.
numeric. Threshold below which a pixel cannot be a tree. Default 2.
numeric. Tolerance see ?EBImage::watershed.
numeric. see ?EBImage::watershed.
logical. By default the functions classify the original point cloud by reference
and return nothing (the original object is automatically updated in place). If
extra = TRUE some additional RasterLayer can be returned.
RasterLayer or data.frame containing the position of the
trees. Can be computed with tree_detection or read from an external file.
numeric. Growing threshold 1. See reference in Dalponte et al. 2016. A pixel is added to a region if its height is greater than the tree height multiplied by this value. It should be between 0 and 1. Default 0.45.
numeric. Growing threshold 2. See reference in Dalponte et al. 2016. A pixel is added to a region if its height is greater than the current mean height of the region multiplied by this value. It should be between 0 and 1. Default 0.55.
numeric. Maximum value of the crown diameter of a detected tree (in pixels). Default 10.
numeric. Maximum value of a crown diameter given as a proportion of the tree height. Default is 0.6, meaning 60% of the tree height.
numeric. For each tree, pixels with an elevation lower than exclusion
multiplied by the tree height will be removed. Thus, this number belongs between 0 and 1.
Nothing, the point cloud is updated by reference. If extra = TRUE
algorithms provide extra outputs. Usually intermediate objects used internally, such as a
RasterLayer.
This method is a growing region method working at the raw point cloud level. It is a
strict implementation of the Li et al. (see references) algorithm made by the lidR
author but with the addition of a parameter hmin to stop the segmentation for objects
that are too low. In practice, this limits over-segmentation when using the method. Otherwise
the algorithm could, for example, segment a lake as a tree.
This is a local maxima + growing region algorithm. It is based on the constraints proposed by
Dalponte and Coomes (see references). This algorithm exists in the package itcSegment.
This version is identical to the original but with superfluous code removed and rewritten
efficiently. Consequently it is hundreds to millions times faster.
Note that this algorithm strictly performs a segmentation, while the original method as
implemented in itcSegment and described in the manuscript also performs a pre-
and post-process when these tasks are expected to be done by the user in separate functions.
This is a simple but elegant method based on local maxima + voronoi tesselation described
in Silva et al. (2016) (see references). This algorithm is implemented in the package
rLiDAR. This version is not the version from rLiDAR. It is
code written from scratch by the lidR author from the original paper and is considerably
(between 250 and 1000 times) faster.
This method is a simple watershed segmentation
method. It is based on the bioconductor package EBIimage.
You need to install this package to run this method (see its github page).
Dalponte, M. and Coomes, D. A. (2016), Tree-centric mapping of forest carbon density from airborne laser scanning and hyperspectral data. Methods Ecol Evol, 7: 1236<U+2013>1245. doi:10.1111/2041-210X.12575. Li, W., Guo, Q., Jakubowski, M. K., & Kelly, M. (2012). A new method for segmenting individual trees from the lidar point cloud. Photogrammetric Engineering & Remote Sensing, 78(1), 75-84. Silva, C. A., Hudak, A. T., Vierling, L. A., Loudermilk, E. L., O<U+2019>Brien, J. J., Hiers, J. K., Khosravipour, A. (2016). Imputation of Individual Longleaf Pine (Pinus palustris Mill.) Tree Attributes from Field and LiDAR Data. Canadian Journal of Remote Sensing, 42(5), 554<U+2013>573. https://doi.org/10.1080/07038992.2016.1196582.
# NOT RUN {
LASfile <- system.file("extdata", "MixedConifer.laz", package="lidR")
las = readLAS(LASfile, select = "xyz", filter = "-drop_z_below 0")
col = pastel.colors(200)
# Li 2012
lastrees(las, "li2012", R = 5)
plot(las, color = "treeID", colorPalette = col)
chm = grid_canopy(las, res = 0.5, subcircle = 0.3)
chm = as.raster(chm)
kernel = matrix(1,3,3)
chm = raster::focal(chm, w = kernel, fun = mean, na.rm = TRUE)
# Dalponte 2016
ttops = tree_detection(chm, 5, 2)
lastrees_dalponte(las, chm, ttops)
plot(las, color = "treeID", colorPalette = col)
# }
Run the code above in your browser using DataCamp Workspace