[Paper Review] AxialGen: A Research Prototype for Automatically Generating the Axial Map

Bin Jiang, Xintao Liu|arXiv (Cornell University)|Feb 3, 2009

Urban Design and Spatial Analysis10 references3 citations

TL;DR

AxialGen is a research prototype that automatically generates axial maps—representations of urban open spaces using the fewest, longest visibility lines (axial lines)—by leveraging a novel 'bucket' concept to group spatial segments based on visibility and connectivity. The method uses medial axis transformations and isovist analysis to identify dominant linear paths, enabling efficient, topology-aware urban space modeling for robotics and urban planning applications.

ABSTRACT

AxialGen is a research prototype for automatically generating the axial map, which consists of the least number of the longest visibility lines (or axial lines) for representing individual linearly stretched parts of open space of an urban environment. Open space is the space between closed spaces such as buildings and street blocks. This paper aims to provide an accessible guide to software AxialGen, and the underlying concepts and ideas. We concentrate on the explanation and illustration of the key concept of bucket: its definition, formation and how it is used in generating the axial map. Keywords: Bucket, visibility, medial axes, axial lines, isovists, axial map

Motivation & Objective

• To develop an automated method for generating axial maps from urban open spaces, reducing manual effort in spatial analysis.
• To address the challenge of representing complex urban layouts with minimal, longest visibility lines that preserve spatial connectivity.
• To introduce and formalize the concept of 'bucket' as a core mechanism for grouping spatial segments based on visibility and geometric continuity.
• To provide a practical, accessible tool for researchers and practitioners in urban planning and robotics to model urban space topology.
• To demonstrate the feasibility of using computational geometry and visibility analysis for scalable urban space representation.

Proposed method

• The method begins by identifying open spaces between buildings and street blocks as the domain for axial map generation.
• It applies medial axis transformation to extract the central skeleton of each open space region.
• The medial axis is segmented into candidate axial lines based on geometric continuity and length.
• A 'bucket' data structure groups these axial lines by visibility and spatial adjacency, ensuring coherence in the final map.
• Visibility analysis using isovists determines which segments are mutually visible, forming the basis for bucket formation.
• The algorithm iteratively merges segments within buckets to form the longest possible axial lines, minimizing the total number of lines in the map.

Experimental results

Research questions

• RQ1 How can axial maps be automatically generated with minimal user input using computational geometry?
• RQ2 What criteria define a 'best' axial map in terms of line count and length for urban space representation?
• RQ3 How can visibility and spatial continuity be algorithmically combined to form coherent axial lines?
• RQ4 What role does the 'bucket' concept play in ensuring topological consistency and minimizing redundancy in axial maps?
• RQ5 Can automated axial map generation achieve comparable quality to manual methods in urban morphology analysis?

Key findings

• The bucket-based approach successfully groups spatial segments into coherent axial lines, reducing the number of lines needed to represent urban open spaces.
• AxialGen generates axial maps with fewer, longer lines by prioritizing visibility and geometric continuity during line formation.
• The method demonstrates robustness in handling complex urban layouts, including irregular and fragmented open spaces.
• The use of isovist analysis enables accurate visibility-based grouping, improving the topological fidelity of the resulting axial maps.
• The prototype proves feasible for real-world urban environments, offering a scalable alternative to manual axial map creation.
• The final axial maps preserve essential spatial relationships, making them suitable for applications in robotics and urban planning.