R Calculate overlapping section (polygon intersection) the fast way

问题

I am looking for a way to calculate the percentage a base-polygon is overlayed by another (selection) polygon.

Take this as an example. I want to calculate the area of red AND blue over the area of red.

group_a <- data.frame(x = c(1, 4, 1),
                      y = c(2, 4, 4),
                      group = "base")
group_b <- data.frame(x = c(2, 5, 2),
                      y = c(3, 3, 5),
                      group = "selection")

dat <- rbind(group_a, group_b)
dat
x y     group
1 1 2      base
2 4 4      base
3 1 4      base
4 2 3 selection
5 5 3 selection
6 2 5 selection

library(ggplot2)

ggplot(dat, aes(x = x, y = y, fill = group)) + geom_polygon(alpha = 0.5)

One way would be to change the data.frames to a Polygon using the rgdal- and rgeos-packages, and then use gArea and gIntersection to find the neccessary values. However, I assume that this might not be the fastest way to do this. Do you know a faster way to calculate the areas/intersections? I would be especially delighted if there is an RCpp-way to do it.

I hope I haven't asked too much; and as always: Thank you for any ideas/hints!

Edit: This is how I would do it with the other packages:

library(rgdal)
library(rgeos)
base <- group_a[, c("x", "y")]
sel <- group_b[, c("x", "y")]

base_pol <- Polygons(list(Polygon(base)), "base")
sel_pol <- Polygons(list(Polygon(sel)), "sel")

shape <- SpatialPolygons(list(base_pol, sel_pol))
plot(shape)

base_area <- gArea(shape["base"])
intersections <- gIntersection(shape["base"], shape["sel"])

gArea(intersections)/base_area
# [1] 0.4027778

Speed Benchmark: Running microbenchmark, I obtained the following results:

OvPerc <- function(base, sel) {
  base <- base[, c("x", "y")]
  sel <- sel[, c("x", "y")]

  base_pol <- Polygons(list(Polygon(base)), "base")
  sel_pol <- Polygons(list(Polygon(sel)), "sel")

  shape <- SpatialPolygons(list(base_pol, sel_pol))
  # plot(shape)

  base_area <- gArea(shape["base"])
  intersections <- gIntersection(shape["base"], shape["sel"])

  return(gArea(intersections)/base_area)
}

library(microbenchmark)
microbenchmark(OvPerc(group_a, group_b), times = 1000)
# Unit: milliseconds
#                    expr      min       lq     mean   median       uq      max neval
# OvPerc(group_a, group_b) 3.680386 3.970394 4.932217 4.157952 4.745398 67.13696  1000

来源：https://stackoverflow.com/questions/35039614/r-calculate-overlapping-section-polygon-intersection-the-fast-way

标签

performance