Cart (Loading....) | Create Account
Close category search window
 

back to article  |  Figures
All Figures

A Global Human Settlement Layer From Optical HR/VHR RS Data: Concept and First Results

Figure 1

Figure 1
Geographic distribution of the HR/VHR input images processed during the experiment.

Figure 2

Figure 2
Spectral coverage of sensors used in the study. The satellites cover a wide spectral range in the visible and NIR part of the spectrum. The spatial coverage includes various resolutions from 50 cm airborne to 10 m panchromatic images of SPOT 2.

Figure 3

Figure 3
The general Discrete Field of Image Descriptors (DFID) concept.

Figure 4

Figure 4
The general IQ GHSL processing workflow.

Figure 5

Figure 5
Cloud detection from panchromatic data: example on a 2.5-m CBERS-2 HRC scene over Brazil.

Figure 6

Figure 6
Example of DAP vector fields: (a) the input image; (b) the color representation of the DAP vector field using the CSL model; (c) the DAP vector field in color-map projection in which the two volumes correspond to the opening top-hat and closing bottom-hat scale-space respectively; (d) a cross-section of the two.

Figure 7

Figure 7
Learning the best HR PANTEX rescaling from low-resolution references. (A) input image, (B) PANTEX feature at DFID 10 m resolution, (C) population LandScan data 1 Km resolution, (D) MODIS Urban data 500 m resolution, (E) PANTEX rescaled according to LandScan (param.set 142), (F) PANTEX rescaled using MODIS500 reference (param.set 146).

Figure 8

Figure 8
Test of the generalization and multi-scale composition options in the city of Sanaa, Yemen. From left to right: i) in white building footprints at 1:10K scale, generalization by ii) dilation, iii) closing, and iv) hybrid approach. White, pink, and green show, respectively, the contribution of the ‘local’, ‘regional’, and ‘global’ scales to the final GHSL product.

Figure 9

Figure 9
Test of the effect in the global representation of the different generalization and multi-scale composition options of the GHSL outputs in the city of Sanaa, Yemen. From left to right: 1) building footprints from cadastral maps 1:10K aggregated to the global GHSL scale using local average; 2) the same using fast ‘nearest-neighbor’ resampling algorithm, 3) using ‘by dilation’, 4) ‘by opening’ and 5) ‘hybrid’ generalization options and also fast ‘nearest-neighbor’ resampling algorithm.

Figure 10

Figure 10
City of Brasilia. top-left: the “presence of buildings” GHSL layer represented at 1:50K scale with the footprints of input scenes (CBERS 2B) in dark-green; top-right: a zoom into the city center. The image shows the “average building size” GHSL layer at 1:10K scale. Increasing BU size is mapped on the blue-green-yellow-red color map; bottom-left and right: the same city represented by the MODIS500 urban layer and BUREF respectively.

Figure 11

Figure 11
Estimated BUREF Agreement of the current GHSL output by WWF eco-region.

Figure 12

Figure 12
Ranking of all the scenes processed during the experiment by increasing Formula$BUREF$ agreement optimized among all the processed options (band, learning parameters) available on the same scene. This is the process implemented for the composition of the final GHSL mosaic, taking the best of the available processed pieces of information. The blue dots represent the worst Formula$BUREF$ agreement available on the same corresponding scenes.

Need Help?


IEEE Advancing Technology for Humanity About IEEE Xplore | Contact | Help | Terms of Use | Nondiscrimination Policy | Site Map | Privacy & Opting Out of Cookies

A not-for-profit organization, IEEE is the world's largest professional association for the advancement of technology.
© Copyright 2014 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.