R - Crop a pixel image produced by spatstat with a boundary shape

Question

i am using the very good spatial statistic package spatstat to produce probability maps of spatial point patterns. Here is a simple working result from the examples to try out.

  require(spatstat)
  data(lansing)
  z <- relrisk(lansing)
  plot(z, main="Lansing Woods")

My question is rather simple (or maybe not). I produced a similar map as shown above and i want to crop it in R with a boundary shapefile. The spatstat package uses its own system (of windows and pixel images) so i am rather clueless. I know that i can format my shapefile to a so called mask with this command as.mask(as.owin(shapefile)), but i didn't get beyond this point.
Anyone knows how i can crop the produced image with this mask?

---

EDIT: Ahh, sry. I already found a solution around it, which works for me. I simply have to make a subset out of the spatial points before converting it to a ppp object.

library(rgdal)
st <- readOGR("GIS-Data","boundary_shape") 
wp <- spp[-which(is.na(over(spp,st)[,1])),] # spp is my spatial point shape
rm(st)
#Then use this command from the maptools package
library(maptools)
plp <- as.ppp.SpatialPointsDataFrame(wp)

Nevertheless it would still be nice to know if it is possible to crop a spatstat pixel image.

Answer 1

The best approach is to coerce your spatstat "image" object to an sp object and use the over function or clip in the raster package. Here is a function to perform the coercion to an sp SpatialGridDataFrame.

as.SpatialGridDataFrame.im <- function(from){
    offset=c(from$xrange[1] + 0.5 * from$xstep, from$yrange[1] + 0.5 * from$ystep)
      cellsize = c(diff(from$xrange)/from$dim[2], diff(from$yrange)/from$dim[1])
dim = c(from$dim[2], from$dim[1])
  gt = GridTopology(offset, cellsize, dim)
    m = t(from$v[nrow(from$v):1, ])
  data = data.frame(v = as.vector(m))
SpatialGridDataFrame(gt, data)
}