[gdal-dev] Hillshade | Raster Band Rendering

[Gareth Grewcock]

Hi - firstly, apologies if the gdal-dev mailing is not the appropriate
mailing list, please advise a more suitable place to post if so.



My objective is to create a hillshaded color-relief image of a DEM using
commandline/programmatic means only, so the process can be automated and
combined within an existing GMT/GDAL workflow.



Currently, I can generate both hillshaded and color-relief rasters using
gdaldem and combine them with Mapnik (using opacity) to generate a my final
rendered image. However, this process will use the min-max values in the
grayscale hillshaded image.



However, I can achieve an improved or sharper hillshaded image when I
manually use QGIS to further process the hillshaded image by:

1. Adding the hillshaded raster to QGIS, which applies by default the
'Cumulative count cut 2% - 98%' style.

2. Then export the hillshaded raster (with the cumulative count style
applied) as a “Rendered Image”.



I have annotated a screenshot here
https://www.dropbox.com/s/ftzk7j2bmznuvjn/raster_band_rendering.png?dl=0
hopefully illustrating the

1.    Improved hillshaded raster output from QGIS ('Cumulative count cut)
compared with,

2.    gdal+mapnik output (using min-max)



So, is there an approach using gdal (or similar command-line tool/app) to
achieve the “improved” hillshaded raster without the “manual” QGIS step?



As I see it, my options are:

1. Use gdalinfo with the “-hist” option to export the histogram of the
hillshaded raster. I guess then I could maybe calculate the 2% and 98%
percentile(?) values and then manipulate the raster values using
gdal_calc.py or something else.  However, I’m no statistician, so hoped
there would be an out of the box solution?!

2. Maybe, I could use the python api for QGIS to import the hillshade,
render, style and export back out. I’m sure this is possible, but would
require an additional python script.

3. Use another library or framework to achieve either of the above. I’ve
researched python’s numpy library, which maybe I could do the percentage
calculation directly on the raster. Again, potentially tricky learning
curve there…



Any help or advice would greatly appreciated. If any help, the data I’m
using is here
https://www.dropbox.com/s/v0peaa3rzaqbhen/raster_band_rendering.zip?dl=0



Cheers


Gareth
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20150318/8f2efb04/attachment.html>