University of Hull logo

A cognitive approach to road recognition with novel feature indicators

Varley, D; Visvalingam, M; Wright, C

Authors

D Varley

M Visvalingam

C Wright

Abstract

This paper outlines a simple strategy for recognition of prototypical cases of urban roads from their outline forms alone. Two new form indicators, namely exits and occupancy, are used together with average width to label regions as roads. They proved to be much more discriminating than the widely used shape metric. Even in its present simplistic form, the proposed strategy for recognition proved to be useful for identifying prototypical cases. The recognition strategy provides an additional means for validating topographic data by cross-checking results previously obtained through the normal process of road extraction. Unlike recognition, extraction relies on manually provided semantic information. Some of the regions which were labelled as roads by the extraction process were flagged as untypical by the recognition process. The flagged regions were mainly those where roads had become fused with neighbouring regions because the linework necessary to separate them, called logical links, had not been manually digitized. Where roads form a relatively small part of these combined regions, the recognition process focuses attention on these by classifying them as untypical.

Journal Article Type Article
Publication Date Sep 1, 1996
Journal The computer journal
Print ISSN 0010-4620
Electronic ISSN 1460-2067
Publisher Oxford University Press (OUP)
Peer Reviewed Peer Reviewed
Volume 39
Issue 9
Pages 751-756
DOI https://doi.org/10.1093/comjnl/39.9.751
Keywords Road recognition
Publisher URL http://comjnl.oxfordjournals.org/content/39/9/751
Additional Information This is a description of an article published in the computer journal, 1996, v.39 issue 9. The full text is not available in this repository.

Downloadable Citations