Adaptive continuity-preserving simplification of street networks

Martin Fleischmann

Charles University

2025-04-24

 

Streets, roads, networks, graphs

Who needs them

geography

urban morphology

cartography

urban economics

transportation planning

spatial demography

segregation

…

representation matters

OpenStreetMap is transport

Araldi and Fusco (2024)

What to do

Manual simplification

click click click click click click click click click click click click click click click click sigh click click click click click click click click click click click click click click click click fml click click click click click click click click click click click click I hate it here click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click click sigh click click click click click click click click click click click click click click click click click help click click click click click click click …

live transcription of the manual simplification process

we need an algorithm

What even is simplification

Illustration of the simplification by the example of a street network fragment in Seraing, Liège (Belgium).

Someone surely has done this

Five requirements

1. open-source
2. fully automated
3. attribute-agnostic
4. case-agnostic
5. fully packaged

Overview of previous studies working towards an automated street network simplification method.

OSMnx node consolidation (Boeing 2025)

cityseer’s simplification workflow (Simmons 2024)

parenx (Voronoi-based) algorithm (Lovelace et al. 2024)

parenx (Voronoi-based) algorithm (Lovelace et al. 2024)

Adaptive continuity-preserving simplification of street networks

What do you mean by “continuity-preserving”

detection of natural continuity of street networks by Tripathy et al. (2020).

Adaptive continuity-preserving simplification of street networks

Conceptual diagram of the proposed simplification algorithm.

Face artifact detection

Fleischmann and Vybornova (2024)

Fleischmann and Vybornova (2024)

Fleischmann and Vybornova (2024)

Conceptual diagram of the proposed simplification algorithm.

Artifact classification

1. artifact polygon contiguity
   1. isolates
   2. pairs
   3. clusters
2. network continuity

Isolates

A subset of the CES typology derived from the Liège street network.

Complete CES typology derived from the Liège street network.

Clusters

merge & skeletonize

Pairs

solve together as a single isolate

solve together as a cluster

Evaluation

numerical assessment

visual assessment

computational performance

selection of case study areas

How close are we to the manually simplified network

Chatterjee’s \(\xi\) correlation coefficient between properties of manually simplified networks and networks based on each of the tested algorithms vs. the original network as a baseline.

Euclidean distance between properties of manually simplified networks and networks based on each of the tested algorithms vs. the original network as a baseline.

A small set of example situations that ought to be simplified and the resutling simplified networks from all tested methods, together with the comparison to the manual simplification considered ground truth.

Output of different simplification methods for a street network fragment in Seraing, Liège (Belgium).

Computational efficiency of different methods.

What works and what does not

Where do I get it

pip install neatnet

Preprint

arxiv.org/abs/2504.16198

Slides

uscuni.org/talks

Code & data

github.com/uscuni/simplification

neatnet

uscuni.org/neatnet