In this paper, we present a probabilistic framework for urban area extraction in remote sensing images using a conditional random field built over an adjacency graph of superpixels. Our discriminative model performs a multi-cue combination by incorporating efficiently color, texture and edge cues. Both local and pairwise feature functions are learned using sharing boosting to obtain a powerful classifier based on feature selection. Urban area are accurately extracted in highly heterogenous satellite images by applying a cluster sampling method, the Swendsen-Wang Cut algorithm. Example results are shown on high resolution SPOT-5 satellite images. © 2009 Springer Berlin Heidelberg.
CITATION STYLE
Besbes, O., Boujemaa, N., & Belhadj, Z. (2009). Cue integration for urban area extraction in remote sensing images. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 5627 LNCS, pp. 248–257). https://doi.org/10.1007/978-3-642-02611-9_25
Mendeley helps you to discover research relevant for your work.