G. F. De Grandi 1 , R.M. Lucas 2 , A. Bouvet 1 European Commission Joint Research Centre 21027, Ispra (VA), Italy e-mail: [email protected]Two-point statistic of polarimetric SAR data provided by a wavelet frame The importance of being scaling…(suggested by Oscar Wilde) Institute of Geography and Earth Sciences Aberystwyth University, Aberystwyth, UK, SY23 3DB. e-mail: [email protected]
16
Embed
TWO-POINT STATISTIC OF POLARIMETRIC SAR DATA TWO-POINT STATISTIC OF POLARIMETRIC SAR DATA
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
G. F. De Grandi1, R.M. Lucas2, A. Bouvet1
European Commission Joint Research Centre21027, Ispra (VA), Italy
What happens when the polarization basis is changed?
vvhvhh SSS 2
Wavelet variance of the crospolar and
copolar power in the rotated polarization
basis
),()(2 fxw
Tvvhvhh SSS 2
),,(),,( '*2
'2 yxyxPxpol
T' 2cos1
2sin
),,(),,( '*1
'1 yxyxPcopol
Wavelet variance of PolSAR power in a rotated basis: a model for a WS stationary process
vvhvhh SSS 2
Hj Fourier transform of the wavelet dilated at scale j
jH
ACF of wavelet coefficients computed in the frequency domain from the
power spectrum G of the input process 21 jinout HGR
outG inG
0Rw out2i Wavelet variance at cross-polarized
state with orientation ψ
Wavelet transformPower
spectrum inPower
spectrum out
Power spectrum of the crosspolar power in the rotated basis
vvhvhh SSS 2 VVHVHH S,S2,S
HHVVHVSScScS
21
Rotation to a linear basis with orientation ψ
Cross-polarized component in the new basis ψ
ii
iab XaP )(
**2 Re)( jijij
iiii
iab SSccSScP
VVHVHHi SSSS ,,
2cos21c
2sin2
22c
Power spectrum of the input process
j
jji
iiin XaXaG *)(
jii XXj
ijiX
iiin
GaaGaG,
2)(
**2 Re)( jijij
iiii
iab SSccSScP
ii
iab XaP )(
Power spectrum in the rotated basis is a linear combination of the power spectra (auto-correlation) and cross-spectra (cross correlation) between dyads of the vector in the H,V basis
Copol-xpol correlation
Power-Correlation
HV2
HH2, VV2
A numerical model for the wavelet variance of a correlated K-distributed stationary clutter
vvhvhh SSS 2
The Wavelet Scaling Polarimetric Signature (WASPS)
),()(2 fxw
Correlated K-distributed clutter, C. Oliver
From theory to practice
vvhvhh SSS 2
Multi-voice wavelet frame
transform
),()(2 sfxWs
VVHVHH SSSfP ,2,
Power synthesis over a range / azimuth transect in a linear
basis with orientation ψ
Supervised wavelet statistics analysis
Single look complex slant range polarimetric data
),()(
)(22
4
sfxW
xW
s
s
Wavelet variance
Wavelet kurtosis (flatness factor)
ALOS PALSAR: Hawai’i island Papau Seamount
vvhvhh SSS 2
Flat sea surface
Rayleigh term
Correlation length
White noise
Data delivered by JAXA ALOS PI program
ALOS PALSAR: Hawai’i island Papau Seamount
vvhvhh SSS 2
Sea surface features
1
3
2
Periodicity
Unbounded
Correlation length
Data delivered by JAXA ALOS PI program
ALOS PALSAR: Hawai’i island Papau Seamount
vvhvhh SSS 2
Sea surface features
1
3
2
Maximum shift
Symmetric signature
Local Maxima
Data delivered by JAXA ALOS PI program
A View from Fourier Kingdom
vvhvhh SSS 2
Spectral Characteristics in the H-V Basis
Sea surface featuresVV VV* HH VV* HH HH*
HH HV* HV VV*
HV HV* Xspec(HH HH*,HH HV*)
Xspec(HV HV*,HH HV*)
Xspec(HV HV*,HH HH*)
Flat sea surface Wind waves
Relative differences in energy normalized by HV power
Power Spectrum copol-xpol correlation (HH HV*): 91%
Cross Power Spectrum (Power, Cross) (HH HH*,HV HH*): 95%
Cross Power Spectrum (Power, Power)(HH HH*,HV HV*): 77%
FLATX WIND
X
*
/
HVHV
XXXWINDFLATWIND
Xspec(Power, Crosscor)
Xspec(Power, Power)
Going to Higher Resolution
DLR Tandem-X dual-pol data
Data provided by DLR AO-2010 VEGE0330
Swamp
Lowland
Mosaic
Clear-cut
Lulonga River – Basankuso - DRC
Going to Higher Resolution
DLR Tandem-X InSAR Coherence
Data provided by DLR AO-2010InSAR processing by SARMAP
Swamp
Lowland
Mosaic
Clearcut
CONCLUSIONS
I will not be the same without the jungleCiao, Leb Wohl, Goodbye, Sayonara