[R] [spatstat] Convert shapefile to pixel image

Rolf Turner r.turner at auckland.ac.nz
Thu Jun 1 23:45:11 CEST 2017

On 02/06/17 01:17, Lluis.Hurtado at uv.es wrote:

> Dear all,
> I am currently working with the spatstat package, using windows and pixel images.
> First:
> My aim is to transform a shapefile (see attached) into a pixel image.
> My idea is to start transforming the shapefile into a Spatial Polygon file:
> x <- readShapeSpatial("200001441.shp")
> y <- as(x, "SpatialPolygons")
> z <- as.owin(y)
> Given z, I want to identify each polygon with a single constant value. This is like adding marks to the SpatialPolygons file. Then I want to convert these polygons into an image, such that the value of each pixel corresponds to the value associated to the polygon where the pixel lies.
> I have been able to do this individually, polygon by polygon, but then I cannot merge the resulting images into a single one. Any idea?
> Second:
> I would also need a single window containing all the smallest polygons (the boundary). I have tried:
> w <- union.owin(z)
> But the resulting window w still shows internal polygons. As read in spatstas FAQ page:
> "First, convert each of the regions into a separate owin object. Then apply union.owin to combine them."
> So I try,
> regions <- slot(y, "polygons")
> regions <- lapply(regions, function(x) { SpatialPolygons(list(x)) })
> windows <- lapply(regions, as.owin)
> But windows is a list of 4307 polygons. How can introduce all of them as a single argument?
>> M <- union.owin(windows)
> Warning messages:
> 1: In union.owin(windows) : Some arguments were not windows
> 2: In union.owin(windows) : No windows were given
> Thank you very much for you help.

As Don has said, this question would be better asked on R-sig-geo.

A *reproducible* example would be nice, e.g. perhaps you could tell us 
how to get the shapefile in question.

In respect of your last point:  Rather than "M <- union.owin(windows)" 
you should use:

     M <- do.call(union.owin, windows)

But that's rather off the track.  What I think you should do (it's hard 
to be certain without a reproducible example) is something like:

tw <- tess(tiles=windows)
iw <- as.im(funxy(as.function(tw,values=vvv),W=Window(tw)),dimyx=128)

where "vvv" is a vector of the values that you wish to associate with 
the individual polygons.  There may be less cryptic ways of 
accomplishing the same thing, but this is the sexiest way, I think.

The value chosen for "dimyx", i.e. 128, is just by way of example.  You 
can use whatever value suits you.  The larger it is, the smoother the 
individual polygons will look, in pixellated form, but the longer things 
will take.  With 4307 polygons, that might be a rather long time!



Rolf Turner

Technical Editor ANZJS
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276

More information about the R-help mailing list