Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2010, Vol. 7, No. 4, pp. 87-96
Comparison of Polarimetric Classification Results Using
ALOS PALSAR Data
M.Y. Dostovalov
, R.V. Ermakov
, T.G. Moussiniants
Research Institute of Precision Instruments, 127490, Dekabristov st. 51, Moscow, Russia
Polarimetric SAR data allow the possibility to increase the amount of information about ground objects obtained
using SAR. ALOS PALSAR data combine polarimetric capabilities with relatively long wavelength. Different methods
of polarimetric classification were tested to detect deforestation areas in the tropical forest using fullpolarization
and dual polarization data. The processing of full-polarization data showed that the intersection of reference
classes degrading the effectiveness of deforestation detection using both supervised and unsupervised methods.
At the same time it was founded that the quality of classification for the same areas improved significantly
using dual polarization data. The main difference between two datasets is the difference in incident angles. Full polarization
datasets were acquired with incident angle ˚2,5 while dual polarization datasets had the corresponding
angle 39˚.
Keywords: SAR, ALOS PALSAR, SAR images, polarimetric classification
Full text