Introduction

RatTree is a tool for building Raster Attribute Tables from a number of thematic images. The output image is created as the intersection of the input images. Pixels which have the same combination of values in all the input images are assigned a common value in the output, and the various input values are carried over as attribute values for that output pixel value. Thus each unique combination of the thematic layers ends up with its own row in the Raster Attribute Tables, allowing more complex processing to happen.

This tool is an example of building more complex data structres in Numba as it creates tree data structures to allow faster searching. There is also a memory saving as only combinations of values found in the input data are stored.

Creating an input CSV

RatTree is both a Python module and a command line utility (rattree_fromimages). We cover the command line here but if you need lower level access please consult the docstrings for the rattree module.

The command line utility takes a path to a .csv file that contains a list of paths to the thematic files and columns names that they are to receive in the Raster Attribute Table.

An example .csv file would look like this:

image_0.img,column 0
image_1.img,column 1
image_2.img,column 2

The resulting RAT will look like this:

Histogram	column 0	column 1	column 2
23341	45	98	12
34553	67	12	78
…	…	…	…

Where the contents of the columns will be the values from the images. Each unique combination of these values gets its own row in the RAT.

Running RatTree from the command line

RatTree needs Numba, GDAL and RIOS installed to run. Having the KEA format installed with GDAL is recommended. Install RatTree with the following command:

pip install git+https://github.com/gillins/rattree.git

Once installed, run rattree_fromimages from the command line. Passing the -h flag gives the usage. Here is in example of how to run it assuming that your .csv file is saved as imagelist.csv:

rattree_fromimages -i imagelist.csv -o result.kea