Generating Road Networks for Old Downtown Areas Based on Crowd-Sourced Vehicle Trajectories

Area of Science:

• Geographic Information Science
• Computer Science
• Transportation Engineering

Background:

• Crowd-sourced trajectory data is increasingly used for road network extraction.
• Extracting road networks in old downtown areas is challenging due to complex layouts, noise, and data sparsity.

Purpose of the Study:

• To propose a novel intersection-first approach for road network generation using low-quality, crowd-sourced vehicle trajectories.
• To address the challenges of complex layouts, high noise, low frequency, and uneven data distribution in old downtown areas.

Main Methods:

• Intersection detection using virtual representative points and the clustering by fast search and find of density peaks (CFDP) algorithm.
• Link extraction employing a Delaunay triangulation network strategy to filter false links.
• Adaptive link-fitting scheme to derive link centerlines, accommodating sparse and uneven data.

Main Results:

• The proposed method significantly improves intersection detection accuracy.
• The approach enhances the quality of road network generation in complex urban environments.
• Experimental results demonstrate superior performance compared to existing methods.

Conclusions:

• The intersection-first approach is effective for road network extraction from challenging crowd-sourced trajectory data.
• The method offers a robust solution for mapping old downtown areas.
• This research contributes to advancements in automated cartography and intelligent transportation systems.