Turn latlngs into image file?

Question

I have a huge CSV file of WGS84 latitudes and longitudes, and I would like to turn these into a large image file that I can print (with each point being a pixel on the image).

What is the best way to do this?

Currently I'm planning to convert them into eastings/northings and then use imagemagick to print a circle at each point, following the approach taken by this author (in the imagemagick script at the bottom of the post).

But I'm wondering if there's a simpler way. I'm not a GIS expert, and perhaps there's even a dedicated geospatial library for doing this?

UPDATE

As requested, here's a sample of the data. The points are all within the UK, but they don't form a regular grid.

Time,latitude,longitude
2014-12-27 18:35:10,51.7438644,-2.2094204
2014-12-27 18:34:08,51.7438657,-2.2094219
2014-12-27 18:33:06,51.7438657,-2.2094219
2014-12-27 18:32:04,51.7438689,-2.2094198
2014-12-27 18:31:03,51.7438716,-2.2094109
2014-12-27 18:30:01,51.7438537,-2.2094211

FURTHER UPDATE

To clarify, I'm looking for a large image that I can print. The points aren't on a regular grid: I was hoping that I could turn each point into a single black pixel on a transparent background.

Answer 1

Using QGIS, the simpliest way is to load the data as delimited text, select EPSG:4326 WGS84 as CRS and save the result with Project -> Save as image.

The Raster -> Conversion -> Rasterize function is a bit more sophisticated, but since your points are close to each other, it needs reprojecting to a projected CRS like UTM zone 10 first to get a reasonable cell size. You need an extra integer column for the cell content as well. Unfortunately, the min/max points are skipped with this method.

Another option is Vector -> Research Tools -> Vector grid around the canvas extent with the option Create as polygons and a cellsize greater 0.1 meters, then run Vector -> Analyze -> Points in polygons on it, and finally Raster -> Conversion -> Rasterize on the polygon layer.

A third chance is to use the SAGA Shape to grid module from the processing toolbox. Keep an eye on the cellsize, the default is too big.

Answer 2

Since the points don't form a regular grid you could use gdal_rasterize.

Set up a VRT header like so:

<OGRVRTDataSource>
    <OGRVRTLayer name="test">
        <SrcDataSource>test.csv</SrcDataSource> 
    <GeometryType>wkbPoint</GeometryType> 
    <GeometryField encoding="PointFromColumns" x="longitude" y="latitude" z="Time"/> 
    </OGRVRTLayer>
</OGRVRTDataSource>

GDAL won't write directly to PNG, but can create them by copying existing rasters, so output to a tiff first.

gdal_rasterize -ot byte -burn 255 -burn 100 -burn 100 -burn 255 -tr 0.01 0.01 -l test test.vrt test.tif

The above command creates a 4 band tiff with the first three bands showing the points in a light red (RGB 255 100 100), you can use any RGB value you want and the fourth band will be used as the PNG transparency (alpha) channel.

You will want to play with the -tr parameters, 0.01 0.01 is roughly 1.1km x 1.1km pixels in WGS84 decimal degrees. You could use the -ts width height parameter instead to specify the number of rows/columns in the output.

Then convert your tiff to a transparent PNG:

gdal_translate -of png -mask 4 test.tif test.png

Answer 3

Here's a fully reproducible example with output:

set.seed(310366) # so we get the same random numbers
library(raster)
uk = getData("GADM",country="GBR", level=0)
bbox(uk) # tells us the bounds (I think it goes as far west as Rockall)
# make 200 points over that area:
pts = cbind(runif(200,-13,1), runif(200,50,60))

That code has done the basic setup.

Now make a 100x100 raster with the approx bounds. Set them right for the grid you want:

r = raster(xmn=-14,xmx=1.8,ymn=49,ymx=61, ncol=100,nrow=100)

If you want transparent, then proceed, otherwise first set all cells to zero with r[]=0.

Now set any cells with a point in them to 1:

r[cellFromXY(r,pts)]=1
plot(r)
plot(uk,add=TRUE)

Of course if you do this with 1000x1000 the pixels will be small and you won't see much unless you have a lot of points.

To create a PNG with transparency, first make a GeoTiff:

writeRaster(r, "uk.tif", format="GTiff", overwrite=TRUE, datatype="INT2U")

Then use ImageMagick "convert" at the shell prompt to create a PNG with white made transparent:

convert -transparent white uk.tif uk.png

Ignore the unknown tag warnings - convert doesn't understand the extra geo-bits of the Tiff. Result, transparent PNG, here zoomed in against a chequerboard:

what more do you want?