Skip to the main content

Original scientific paper

Super Resolution Mapping of Agricultural Parcel Boundaries based on Localized Partial Unmixing

Dragutin Protić ; Faculty of Civil Engineering, University of Belgrade, Belgrade, Serbia
Milan Kilibarda ; Faculty of Civil Engineering, University of Belgrade, Belgrade, Serbia
Ivan Nestorov ; Faculty of Civil Engineering, University of Belgrade, Belgrade, Serbia


Full text: english pdf 1.836 Kb

page 259-271

downloads: 435

cite


Abstract

Operational mapping of crop/agricultural parcel boundaries is a challenging task. It should be efficient and inexpensive process resulting with data useful for wide range of purposes. Hard classification of medium resolution (>15 m, <30 m) remote sensing data could be a solution but the spatial resolution of the imagery is not sufficient for many applications. In this paper a method for soft classification of medium remote sensing data and super-resolution mapping of crop boundaries is presented. The method is developed based on localized partial unmixing with estimating both target and background end-member spectra. The output was subjected to pixel swapping as super resolution mapping technique. Testing is done on real Landsat ETM+ imagery. The method is evaluated quantitatively and the results show significant increase of accuracy when compared to mapping from original remote sensing data.

Keywords

agricultural parcel mapping; super resolution mapping; partial unmixing; linear unmixing; pixel swapping

Hrčak ID:

102914

URI

https://hrcak.srce.hr/102914

Publication date:

21.12.2012.

Article data in other languages: croatian

Visits: 972 *