ABSTRACT
Increasing demands for up-to-date road network and the availability of very-high-resolution (VHR) satellite images as well as the popularity of high-speed computers provide motivation and preliminary materials for researchers to propose more advanced approaches in order to increase the automation and robustness of road extraction strategies. In this article, road characteristics are modelled via object-based image analysis (OBIA). Object-based information is embedded as heuristic information in the ant colony optimization (ACO) algorithm for handling the road network extraction problem. A new neighbourhood definition in object space is introduced, which affects the transition rule in order to decrease the road gaps. Furthermore, an innovative desirability function for ACO is designed, which extracts the road objects, competently. The experimental results demonstrate the efficiency of the proposed algorithm for road extraction from VHR images. Moreover, the results of two state-of-the-art methods are compared with the proposed algorithm.
Acknowledgements
The authors would like to thank the ISPRS and the Space Imaging LLC for providing the IKONOS data set. The authors would also like to thank Professor Timothy Warner and anonymous reviewers for their constructive suggestions, which have enriched this article.
Disclosure statement
No potential conflict of interest was reported by the authors.