parenx

Simplify (or "pare") a GeoJSON network ("nx") using raster image skeletonization an Voronoi polygons

Provides functions that use image skeletonization or Voronoi polygons to simplify geographic networks composed of linestrings. The outputs are geographic layers representing simplified or 'primal' representations of the network. Primal networks only contains straight line segments

Sample datasets include:

• Princes Street in Edinburgh, in data/rnet_princes_street.geojson
• Rail lines in Doncaster, in data/rnet_doncaster_rail.geojson

<!-- Todo: add more -->

Installation

Install the package into an activated python virtual environment with the following command:

pip install parenx

Install the latest development version from GitHub with the following command:

pip install git+https://github.com/anisotropi4/parenx.git

This places the skeletonization.py and voronoi.py scripts into the executable search path.

Test to see if the package is installed with the following command:

python -c "import parenx; print(parenx.__version__)"

Examples

A bash helper script run.sh and example data is available under the sitepackages/parenx project directory under venv. The exact path varies with module and python version

Example data

# Download the data if not already present
if [ ! -f ./data/rnet_princes_street.geojson ]; then
    wget https://raw.githubusercontent.com/anisotropi4/parenx/main/data/rnet_princes_street.geojson
    # Create data folder if not already present
    if [ ! -d ./data ]; then
        mkdir ./data
    fi
    mv rnet_princes_street.geojson ./data
fi

Skeletonization

The following creates a simplified network by applying skeletonization to a buffered raster array in output.gpkg

<!-- (venv) $ ./skeletonize.py data/rnet_princes_street.geojson -->

skeletonize.py ./data/rnet_princes_street.geojson rnet_princes_street_skeletonized.gpkg

tile_skeletonize.py ./data/rnet_princes_street.geojson rnet_princes_street_skeletonized_tile.gpkg

Voronoi

The following creates a simplified network by creating set of Voronoi polygons from points on the buffer in output.gpkg

<!-- (venv) $ ./voronoi.py data/rnet_princes_street.geojson -->

voronoi.py ./data/rnet_princes_street.geojson rnet_princes_street_voronoi.gpkg

Simple operation

The run.sh script sets a python virtual environment and executes the script against a data file in the data directory

$ ./run.sh

The run.sh script optionally takes a filename and file-extension. To simplify a file, say somewhere.geojson and output to GeoPKG files sk-simple.gpkg and vr-simple.gpkg

$ ./run.sh somewhere.geojon simple

Locating the run.sh script

To copy the run.sh script into your local directory the following could help

$ find . -name run.sh -exec cp {} . \;

Using the parenx helper script

A dash helper script parenx is also available under the sitepackages/parenx project directory under venv. The exact path varies with module and python version

Locating the parenx script

To copy the parenx script into your local directory the following could help

$ find . -name parenx -type f -exec cp {} . \;

Simplification using different algorithms

The parenx helper script allows the algorithm to be selected as a command line parameter for the three supported algorithms:

./parenx skeletonize ./data/rnet_princes_street.geojson rnet_princes_street_skeltonize.gpkg
./parenx tile_skeletonize ./data/rnet_princes_street.geojson rnet_princes_street_tile.gpkg
./parenx voronoi ./data/rnet_princes_street.geojson rnet_princes_street_voronoi.gpkg

Application Programming Interface (API) Example

The skeletonize_frame, voronoi_frame, primal_frame and tile_skeletonize_frame functions are exposed via a simple API.

#!/usr/bin/env python3

import geopandas as gp
from parenx import skeletonize_frame, voronoi_frame, skeletonize_tiles, get_primal

CRS = "EPSG:27700"
filepath = "data/rnet_princes_street.geojson"
frame = gp.read_file(filepath).to_crs(CRS)

parameter = {"simplify": 0.0, "buffer": 8.0, "scale": 1.0, "knot": False, "segment": False}
r = skeletonize_frame(frame["geometry"], parameter)

parameter = {"simplify": 0.0, "scale": 5.0, "buffer": 8.0, "tolerance": 1.0}
r = voronoi_frame(frame["geometry"], parameter)

primal = get_primal(r)

Notes

Both are the skeletonization and Voronoi approach are generic approaches, with the following known issues:

• This does not maintain a link between attributes and the simplified network
• This does not identify a subset of edges that need simplification
• The lines are a bit wobbly
• It is quite slow