AN IMPROVEMENT ON GEOMETRY-BASED METHODS FOR GENERATION OF NETWORK PATHS FROM POINTS

Akbari, Z.; Abbaspour, R. A. · 2014 · DOAJ

DOI: 10.5194/isprsarchives-XL-2-W3-13-2014

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 challenge of converting discrete point data from positioning systems, such as GPS, into accurate network paths. While GPS data is widely used for applications like traffic analysis and speed estimation, the inherent errors in positioning often result in points that do not align exactly with road centerlines. The authors aim to improve geometry-based methods for estimating network paths from these points, specifically focusing on mitigating errors that occur at intersections and in high-density network areas. The study evaluates three existing geometry-based methods: the Direct Line Method, which connects points with straight lines; the Point-to-Point Method, which projects observed points onto the closest network vertices; and the Point-to-Curve Method, which maps points to the closest network arcs. The authors assess these methods by comparing the number of common arcs and the total length of estimated paths against actual paths for three test cases. Results indicate that the Direct Line Method performs poorly, with zero overlap with actual network arcs. The Point-to-Point Method often overestimates path length and fails to encompass all actual lines. The Point-to-Curve Method yields the highest accuracy but suffers from specific weaknesses, including incorrect estimations at intersections, ignoring lack of connections between arcs, and errors in high-density networks. To address these limitations, the authors propose an improved algorithm based on the Point-to-Curve Method. This method introduces specific geometric conditions to filter potential paths. First, a buffer zone is created around the path estimated by the Direct Line Method; only network lines within this buffer are considered. Second, if the closest line between two consecutive points lies on a single arc, the intermediate point must also be mapped to that same arc. Third, the algorithm prevents the repetition of consecutive lines in a reciprocating manner, which often results from intersection errors. The optimal buffer distance is determined by selecting the value that minimizes the difference between the estimated path length and the direct line length. The proposed method was tested against the same three paths used for the initial evaluation. The results demonstrate significant improvement, with the estimated paths matching the actual paths in terms of both arc count and length for two of the three test cases, and showing high accuracy for the third. The authors conclude that this fully geometric approach effectively minimizes path derivation and errors associated with intersections and dense networks, offering a simpler alternative to complex mathematical or statistical map-matching algorithms.

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.

• exposure measurement