Identification of Curves and Straight Sections on Road Networks from Digital Vector Data

Andrášik, Richard; Bíl, Michal; Janoška, Z.; Valentová, Veronika · 2013 · OpenAlex-citations

DOI: 10.2478/v10158-012-0033-0

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the need for an automated method to distinguish between horizontal curves and straight sections on road networks using digital vector data. The motivation stems from traffic safety concerns, as accident rates in curves are significantly higher than on straight sections due to lane departure and visibility issues. Additionally, identifying these geometric features aids road administrators in managing infrastructure, such as installing crash barriers, and helps analyze driver behavior, particularly the calming effect of curves versus the monotony of long straight sections. The authors propose a method based on calculating the radius of an osculating circle at each point along a road line, which approximates the local curvature. To ensure data quality, the raw geographic data—sourced from the Road and Motorway Directorate of the Czech Republic—was first pre-processed using the Douglas-Peucker algorithm for geometric line generalization. This step reduces redundant points and simplifies the line while maintaining essential geometric information. The proposed osculating circle method was compared against three alternative approaches: calculating the radius of a circumscribed circle through three consecutive points, measuring the angle formed by three consecutive points, and calculating the cumulative angle of three points. The methodology was tested on two road sections in the Olomouc district: an 8.7 km segment of road II/446 with shallow curves and a 5.6 km segment of road II/444 with frequent horizontal and vertical curves. A traffic engineering expert manually classified each point as either a curve or a straight section to serve as a ground truth for validation. The results demonstrated that methods based on angle calculations were unsuitable for clear classification, as their value distributions did not separate curves from straight sections effectively. In contrast, radius-based methods performed significantly better. Specifically, the osculating circle method outperformed the circumscribed circle method in identifying curves. For instance, on section II/444 with a 1000 m radius threshold, the osculating circle method achieved a specificity of 83.8% and a positive predictive value of 95.9%, compared to 79.5% and 94.5% respectively for the circumscribed circle method. Both methods showed similar sensitivity for identifying straight sections. The study concludes that the osculating circle method is the most effective for automatically identifying curves and straight sections from digital vector data. It is robust against data irregularities that cause false positives in circumscribed circle calculations and is easily implementable in Geographic Information Systems (GIS). This tool facilitates broader use by safety experts and road administrators for network analysis and management, though the determination of optimal classification thresholds remains an area for further research.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• road geometry