How to get the centerline of a polygon without artefacts in Python?

Question

I am using Python to obtain the central line of a polygon (blue line) which represents the contour of a road (red line). To do so, I have used the Centerline library as shown below:

from centerline.geometry import Centerline
import geopandas as gpd

# Load the polygon representing the road contour as a GeoDataFrame

shapefile_path = 'contorno_alpha02_suavizado025_cc.shp'
gdf = gpd.read_file(shapefile_path)

# Calculates central line of the polygon

gdf['centerline'] = gdf['geometry'].apply(
    lambda x: calculate_centerline(x, {"id": 1, "name": "polygon", "valid": True}))

# New GeoDataFrame to store central lines

centerlines_gdf = gpd.GeoDataFrame(geometry=gdf['centerline'])

# Save central lines in a new shapefile.

output_shapefile_path = 'centerlines_shapefile_025_CC.shp'  
centerlines_gdf.to_file(output_shapefile_path)

The code works and gives a central line, but it has several branches that are not desirable, as they are not part of the road axis. The branches appear mainly in the curved parts of the polygon (see attached images). Is there any way to removes those "branches" with Python (ideally) or in any GIS software?

Answer 1

The pygeoops library has a functionality that tries to do exactly that: pygeoops.centerline. There are some optional parameters you can use to finetune the generation and cleanup if needed.

Sample script based on your code (untested as I don't have a test file):

import geopandas as gpd
import pygeoops

# Load the polygon representing the road contour as a GeoDataFrame
shapefile_path = 'contorno_alpha02_suavizado025_cc.shp'
gdf = gpd.read_file(shapefile_path)

# Calculate centerline of the polygons
gdf.geometry = pygeoops.centerline(gdf.geometry)

# Save centerlines in a new shapefile.
output_shapefile_path = 'centerlines_shapefile_025_CC.shp'  
gdf.to_file(output_shapefile_path)

Disclaimer: I'm the developer of pygeoops.