7yl4r / imagery_utils Goto Github PK

PGC Imagery Utils


INTRODUCTION
------------
PGC Imagery Utils is a collection of commercial satellite imagery manipulation tools to handle bathch processing of a few tasks:

1) Correct for terrain and radiometry
2) Mosaic several images into one set of tiles
3) Pansharpen a multispectral image with its panchromatic partner
4) Calculate an NDVI raster from a multispectral image.

These tools are build on the GDAL/OGR image processing API using Python.  The code is built to run primarily on a Linux HPC cluster running Maui/Torque for queue management.  Some of the tools will work on a Windows platform.

The code is tightly coupled to the systems on which it was developed.  You should have no expectation of it running on another system without some patching.

SCRIPT USAGE
------------
Files starting with "qsub" and "slurm" are PBS and SLURM submission scripts.  See the script-specific documenation for more details on usage.

ORTHO

The orthorectification script can correct for terrain displacement and radiometric settings as well as alter the bit depth of the imagery.  Using the --pbs or --slurm options will sumit the jobs to an Torque job sceduler.  Alternatively, using the --parallel-proceses option  will instruct the script to run multiple tasks in parallel.

Example:  python pgc_ortho.py --epsg 3031 --dem DEM.tif --format GTiff --stretch ns --outtype UInt16 input_dir output dir

This example will take all the nitf or tif files in the input_dir and orthorectify them using DEM.tif.  The output files will be written to outdut_dir and be 16 bit (like the original image) GeoTiffs with no stretch applied with a spatial reference of EPSG 3031, or Antarctic Polar Stereographic -71.  

MOSAIC

The mosaicking toolset mosaics multiple input images into a set of non-overlapping output tile images.  It can sort the images accounding to several factors including cloud cover, sun elevation engle, off-nadir angle, probability of overexposure, and proximity to a specific date.  It consists of 3 scripts:

1. pgc_mosaic.py - initializes the output mosaic, creates cutlines, and run the subtile processes.
2. pgc_mosaic_query_index.py - takes mosaic parameters and a shapefile index and determines which images will contribute to the resulting mosaic. The resulting list can be used to reduce the number of images that are run through the orthorectification script to those that will be eventually used.
3. pgc_mosaic_build_tile.py - builds an individual mosaic tile.  This script is invoked by pgc_mosaic.

Example:  python pgc_mosaic.py --pbs --bands 1 --tilesize 20000 20000 --resolution 0.5 0.5 input_dir output_mosaic_name

This example will evalutate all the 1-band images in input_dir and sort them according to their quality score.  It will submit a job to the cluster queue to build each tile of size 20,000 x 20,000 pixels at 0.5 meters resolution.  The output tiles will be Geotiffs named by appending a row and column identifier to the output_mosaic_name.

PANSHARPEN

The pansharpening utility applies the orthorectification process to both the pan and multi image in a pair and then pansharmens them using the GDAL tool gdal_pansharpen.  GDAL 2.1 is required for this tool to function.

NDVI

The NDVI utility calculates NDVI from multispectral image(s).  The tool is designed to rn on data that have already been run through the pgc_ortho utility.

INSTALLATION AND DEPENDANCIES
-----------------------------
LINUX

Installation of PGC's GDAL stack is documented on our Github page (https://github.com/PolarGeospatialCenter/gdal-full).

WINDOWS

The OSGeo4w Express Installation provides a Windows enviroment to use some of these tools.  You can get the installers here: http://trac.osgeo.org/osgeo4w/.


CONTACT
-------
Claire Porter
Polar Geospatial Center
[email protected]