Road Map Inference: A Segmentation and Grouping Framework

We propose a new segmentation and grouping framework for road map inference from GPS traces. We first present a progressive Density-Based Spatial Clustering of Application with Noise (DBSCAN) algorithm with an orientation constraint to partition the whole point set of the traces into clusters that r...

Full description

Saved in:
Bibliographic Details
Published in:ISPRS international journal of geo-information 2016-08, Vol.5 (8), p.130-130
Main Authors: Qiu, Jia, Wang, Ruisheng
Format: Article
Language:English
Subjects:
2D point cloud segmentation
Algorithms
Clusters
Construction
DBSCAN
HMM map matching
Inference
map inference
Markov chains
point cluster grouping
Proximity
Roads
Segmentation
Topology
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We propose a new segmentation and grouping framework for road map inference from GPS traces. We first present a progressive Density-Based Spatial Clustering of Application with Noise (DBSCAN) algorithm with an orientation constraint to partition the whole point set of the traces into clusters that represent road segments. A new point cluster grouping algorithm, according to the topological relationship and spatial proximity of the point clusters to recover the road network, is then developed. After generating the point clusters, the robust Locally-Weighted Scatterplot Smooth (Lowess) method is used to extract their centerlines. We then propose to build the topological relationship of the centerlines by a Hidden Markov Model (HMM)-based map matching algorithm; and to assess whether the spatial proximity between point clusters by assuming the distances from the points to the centerline comply with a Gaussian distribution. Finally, the point clusters are grouped according to their topological relationship and spatial proximity to form strokes for recovering the road map. Experimental results show that our algorithm is robust to noise and varied sampling rates. The generated road maps show high geometric accuracy.
ISSN:2220-9964
2220-9964
DOI:10.3390/ijgi5080130