GEOSCAN Search Results: Fastlink

GEOSCAN Menu


TitleUrban land use mapping using high resolution SAR data based on density analysis and contextual information
AuthorChen, Z; Zhang, Y; Guindon, B; Esch, T; Roth, A; Shang, J
SourceCanadian Journal of Remote Sensing vol. 38, no. 6, 2012 p. 738-749, https://doi.org/10.5589/m13-002 (Open Access)
Year2012
Alt SeriesEarth Sciences Sector, Contribution Series 20120394
PublisherInforma UK Limited
Documentserial
Lang.English
Mediapaper; on-line; digital
File formatpdf
Subjectsgeophysics; remote sensing; mapping techniques; computer mapping; satellite imagery; land use; SAR
Illustrationstables; flow charts; aerial photographs; satellite images
ProgramMethodology, Remote Sensing Science
Released2014 06 04
AbstractThis paper presents a procedure for urban land use interpretation from a single high-resolution synthetic aperture radar (SAR) image. The approach involves two semi-automatic steps: urban extent delineation and urban land use mapping. In the first step, two general classes (urban and nonurban) are mapped using an existing method that involves analysis of speckle characteristics and intensity information. In the second step, more detailed urban land use classification is undertaken based on analysis of regional radar backscatter patterns in terms of density of dark linear features, density of bright features, and urban contextual information. Density analysis was conducted at three levels: individual building\'02road, urban block, and suburban commercial\'02industrial. Contextual information, including density, building size, and distance between buildings and parking places, was used to quantify urban morphological patterns. Tests were conducted for mapping Ottawa, Canada, using five Radarsat-2 images of different incidence angles and three TerraSAR-X images of the same incidence angles but different dates. The results show that the proposed method could be used to map five urban land uses including low-density residential, commercial\'02industrial, high-density urban, open land, and nonurban with accuracies in the range from 74% to 82%.
GEOSCAN ID292218