Sub-pixel mapping of rural land cover features from fine spatial resolution remotely sensed imagery

Abstract

Mapping rural land cover features, such as trees and hedgerows, for ecological applications is a desirable component of the creation of cartographic maps by the Ordnance Survey and inclusion in geographic database systems such as OS Mastermap®.  Based on the phenomenon of spatial dependence, sub-pixel mapping can provide increased mapping accuracy of such features.  A simple pixel swapping algorithm for super-resolution sub-pixel mapping was applied to predicted class proportions derived from a soft classification of simulated and real fine spatial resolution remotely sensed imagery.  Input proportions were super-resolved into sub-pixels using a specified zoom factor.  Sub-pixels were then iteratively swapped until the spatial correlation between sub-pixels for the entire image was maximised. The standard pixel swapping algorithm was developed to increase the accuracy with which rural land cover features were predicted.  Firstly, the algorithm was modified to increase the likelihood of predicting linear features, such as hedgerows, on the basis of measured anisotropy within class proportions.  Secondly, an image fusion component was integrated to enable the use of multiple datasets such as panchromatic imagery, to refine the prediction of the geometric characteristics of the predicted features.  The new pixel swapping technique increased the accuracy with which rural land cover features compared with the standard technique and substantially increased the utility of such land cover maps compared with conventional mapping techniques, such as classification.

More information

Identifiers

Catalogue record

Export record

ASCII CitationAtomBibTeXData Cite XMLDublin CoreDublin CoreEP3 XMLEndNoteHTML CitationHTML CitationHTML ListJSONMETSMODSMPEG-21 DIDLOpenURL ContextObjectOpenURL ContextObject in SpanRDF+N-TriplesRDF+N3RDF+XMLRIOXX2 XMLReferReference ManagerSimple Metadata