Horizontally slicing point cloud into a cylinder

Question

I have some point clouds from mobile terrestrial laser scanning of circular sampling plots in woodland environments. I am interested in studying the structural complexity of the woodland understorey. I have used the clip_circle() function to crop the point clouds to the 15 metre radius of the sampling plot.

Now, I would like to horizontally slice the point cloud so that I can obtain the first two metres of points from the woodland floor. Therefore, I would like to extract a 15-metre radius cylinder of two metres depth, positioned on the floor.

I managed to do this in CloudCompare and reimport the cloud back into R like this:

However I would like to do the whole pipeline in R to save time when I have more clouds to process.

Answer 1

There are several options to achieve this task (assuming the point cloud is height normalized)

The simplest options is to read only the point below 2 meters and then clip

las = readLAS(file, filter = "-drop_z_above 2")
plot = clip_circle(las, x, y, radius)

On other option is to do the same from a LAScatalog. This is more suitable for multiple queries an more memory optimized if the original file is particularly big

ctg = readLAScatalog(file, filter = "-drop_z_above 2")
plot = clip_circle(ctg, x, y, radius)

The last option and the most trivial but the least efficient is to filter the points of interest with filter_poi()

las = readLAS(file)
las = filter_poi(las, Z <= 2)
plot = clip_circle(las, x, y, radius)

Answer 2

It's not memory-safe like catalog_apply, but you could process scans in parallel as shown below.

 #install packages and libs w/ pacman ----------------------------------------
if (!require("pacman")) install.packages("pacman"); library(pacman)
p_load(lidR, TreeLS, doParallel, parallel, here)

# read files into a list of data frames ------------------------------------
dat = as.data.frame(list.files(here("path/to/normalized/data"), pattern="laz$"))
seq_id_all = seq_along(1:nrow(dat))

# define function -----------------------------------------------------------
my_fun <- function(...) {
 i <- (...)
# load all libs into function
library(TreeLS); library(lidR); library(here); library(tidyverse)  
zeb_data <- readTLS(here("path/to/normalized/data", dat[i,]))
(a <- zeb_data %>% filter_poi(Z >= 0 & Z <1)) # filter by Z (0 to 1m)
(b <- zeb_data %>% filter_poi(Z >= 1 & Z <2))

# write output
writeLAS(a, file= here("output/path", paste0(dat[i,], "_0to1m.laz")))
writeLAS(b, file= here("output/path", paste0(dat[i,], "_1to2m.laz")))
}

# get number of available cores for parallel processing ------------
n_cores = detectCores(logical = TRUE)
cl = makeCluster(n_cores-1)  # leave 1 core availible
registerDoParallel(cl)  # register the cluster

# export function and 'dat' to the cluster and run ----------------
clusterExport(cl, list('my_fun','dat'))
# create a output directory
dir.create(here('output/path'))
system.time(results <- c(parLapply(cl, seq_id_all, fun=my_fun)))
stopCluster(cl)