How to create a lowest function for decimate_points() using lidR

Question

How do I create a custom algorithm for decimate_points() in lidR analogous to the highest() algorithm but that selects the lowest points?

decimate_points(las, lowest(2))

I know that lidR allows to add custom algorithms as it was suggested by @JRR in this question: How -thin_with_grid works?

Answer 1

lidR accepts "plugins". How to make plugins is partially documented in a dedicated chapter of the book. It is not yet fully part of public API but you can safely use it if you are not creating a CRAN package.

Notice that a C++ based version of lowest() will be available in lidR 3.1.2

library(lidR)

lowest <- function(res = 1) {
  stopifnot(is.numeric(res), length(res) == 1L, res > 0)
  
  f <- function(las) {
    r <- grid_metrics(las, ~.I[which.min(Z)], res  = res) # .I[] is a data.table feature
    return(na.omit(r[])) # Return the id of the points of interest
  }
  
  class(f) <- lidR:::LIDRALGORITHMDEC # Using ::: because not public yet
  return(f)
}

lowest(2)
#> Object of class lidR algorithm
#> Algorithm for: point cloud thinning 
#> Designed to be used with: decimate_points 
#> Native C++ parallelization: no 
#> Parameters: 
#>  - res = 2 <numeric>

LASfile <- system.file("extdata", "Megaplot.laz", package="lidR")
las <- readLAS(LASfile)
thinned <- decimate_points(las, lowest(4))
plot(thinned)