www.jrc.ec.europa.eu Serving society Stimulating innovation Supporting legislation Geographical Information Systems (GIS) orthoimagery Wim Devos Wim Devos
Mar 27, 2015
www.jrc.ec.europa.eu
Serving societyStimulating innovationSupporting legislation
Geographical Information Systems (GIS) orthoimagery
Wim Devos
Wim Devos
Outline
• Resolution• Sensor types• Geometry: Orthorectification• Radiometry:
Multispectral/Panchromatic/Pansharpened• Common CAP sensor/platforms
Spec
tral
Res
olut
ion
Topographic Mapping
Transportation
Defense
Urban
AgricultureEarth Resources
Environment
Forestry
100 m 10 m 1 m 0.1 m 0.01 m
Spatial Resolution
Panc
hrom
atic
M
ultis
pect
ral
H
yper
spec
tral
Sensor application segments (2004)
LPIS ?
Airborne Sensors
Airborne vs Spaceborne capture
12 km / 40,000 ft 1.3
million
ft
to
2
.5 m
illion
ft
1 km / 3,000 ft
Spaceborne Sensors
800 km / 500 miles
400 km / 250 miles
Spaceborne GSD > 0.80m
Airborne Digital GSD 0.20m
Airborne Film GSD 0.10m
400,0
00 m
t
o
760,0
00 m
OBSOLETE in 2013!
Complementary
Data on demand Can operate in adverse
weather conditions( flying under clouds )
Adaptable resolution0.1 - 0.8 m Pan0.2 - 1.6 m Multispectralby changing flying height
Stereo imagery is inherent
Spaceborne sensors (Hi-Resolution)Airborne digital sensors
Fixed orbit (450-650 km) Availability is weather
dependent Fixed resolution
0.8 m Pan4.0 m Multispectral
Known cost per scene Stereo on demand
12.8m 6.4m. 3.2m 1.6m
0.80m 0.40m 0.20m 0.10m
Resolution (pixel size, ground sampling distance)≡ dimension on the ground
GSD 1.6m
GSD 0.20m
Resolution and detection-interpretation
Size of recognizable objectGSD x 3
Car size ~ 4.5m - 5m GSD 1.6 m x 3 = 4.8 m
Size of interpretable objectGSD x 21
Car size ~ 4.5 - 5m GSD 0.2 m x 21 = 4.8 m
the nature of the object determines the resolution required!
Orthorectification≡ process of removing perspective and terrain distortion
using a Digital Elevation Model(DEM)
Result = constant scale
In practice
raw frame orthorectified frame
mosaic
Possible orthorectification issuesGeometric correction• SRS issues• Inappropriate Ground Control Points• GSD-pixel ratio• DEM weaknessesRadiometry• Inappropriate re-sampling algorithm of GS-DN• Mosaicking• Histogram (saturation, alteration,...)• Processing artefactsGuidance (for LPIS):
http://marswiki.jrc.ec.europa.eu/wikicap/index.php/Orthoimage_technical_specifications_for_the_purpose_of_LPIS
example original capture produced
orthoimage
50cm GSD turned into 1m pixel
example correct image CRS incorrect image
CRS “rough” DEM “smoother” result but
why?
despite looks, vector has true location!
Coordinates are correct at ground level!
• treetop line – walls (consider the basis)
• ditches - depressions (consider the top)• steep slopes - embankments (take into account)
Why hyperspectral?
Atmospheric scattering and absorption• The sky is blue because....• Easy sun tan in the mountains because… longer wavelenghts provide a sharper image (contrast)
Reflective behaviour of vegetation• turgor (H20) NIR reflection• chlorofyl VIS absorption discrimination enhanced
False colour = better contentwater
This tiny peak makes a leaf appear green! Eyes don’t pick up its total NIR reflection/transparency!
false colour G-R-NIR BGR
NIR Red
GreenColor Composite image in RGB space generated by combining images taken at different wavelengths
Panchromatic band
broader band, more light smaller GSD
NIR
panchromatic multispectral
= bundled product
Pansharpening combines PAN+MS:
inappropriate appropriate
Synthetic aperture radars
Oblique (side looking)+: 24/7 (active system)
- : signal strength = f(geometry, roughness, polarisation, dielectric properties)
C-band fully polarimetric image composed by the dual pol DLR’s E-SAR system. The intensity image show the difference on the agricultural fields expressed in different colors. The colors are derived by the combination of different polarisations (red=HH, blue=VV, green=HV)
Radar image
Conclusion
Ortho-rectification removes perspective and displacement to create a metric canvas
• suitability of image sources depends on spatial and radiometric properties of the features
• production by off-the-shelf technology but guidance needs to be followed to prevent loss of information
• correct interpretation is criticalthere are many CAPI techniques not discussed here
• more imagery always provides more information (phenology, changes)
• INSPIRE annex II theme: free use within public services should be assumed