tile_at_coords: Generate tiles from imagery and spatial data

In mosscoder/paint2train: What the Package Does (One Line, Title Case)

Description

Accepts a two-column matrix of coordinates and a path to a multi-band raster. Generates tiles of specified side length in CRS units of the data with an optional buffer to avoid edge effects in subsequent preocessing. Save tiles to an output directory for further preprocessing.

Usage

tile_at_coords(coords, img, len_side, out_dir, buffer = 0, ncores = 1)

Arguments

coords	A two column matrix containing x and y coordinates.
img	The path to the source raster stack.
len_side	Desired side length of tiles in img CRS units.
out_dir	Directory in which to place tile output.
buffer	Distance in native CRS units to buffer around tiles.
ncores	The number of cores to use for parallel tiling.

Value

None. Tiles are saved to out_dir.

Examples

library(paint2train)

image_dir <- tempfile()
image_url <- 'https://storage.googleapis.com/mpgranch_data/sample_4band.tif'
download.file(url = image_url, destfile = image_dir)
tdir <- tempdir()
setwd(tdir) 
preproc_dir <- 'preproc_tiles'
dir.create(preproc_dir)

#some test coordinates
xcoords <- c(727495,
             727919)

ycoords <- c(5175339,
             5175408)

coord_mat <- cbind(xcoords, ycoords)

ls <- 30 #how big should the tiles be, this is the side length (in units of data, meters here)
buff <- 5  #buffer in native units of CRS
cores <- ifelse(.Platform$OS.type == 'unix', #how many cores to use for preprocessing
                   parallel::detectCores() - 1,
                   1) 
                   
tile_at_coords(coords = coord_mat,
 len_side = ls,
 buffer = buff,
 out_dir = preproc_dir,
 img = image_dir,
 ncores = cores)
 
#show tiling outcome
par(mfrow = c(1,2))              
for(i in seq_along(xcoords)){plotRGB(stack(list.files(preproc_dir, full.names = T)[i])[[1:3]])}
par(mfrow = c(1,1))
 

mosscoder/paint2train documentation built on Jan. 21, 2022, 11 a.m.