Classification of full-polarization ALOS-PALSAR imagery using SVM in arid area of Dunhuang

Classification is an important process in interpretation of synthetic aperture radar （SAR） imagery. As an advanced instrument for remote sensing, the polarimetric SAR has been applied widely in many fields. The main aim of this paper is to explore the ability of the full-polarization SAR data in classification. The study area is a part of Dunhuang, Gansu Province, China. An L-band full-polarization image of Dunhuang which includes quad-polarization modes was acquired by the ALOS-PALSAR （Advanced Land Observing Satellite-the Phased Array type L-band Synthetic Aperture Radar）. Firstly, new characteristic information was extracted by the difference operation, ratio operation, and principal component transform based on the full-polarization （HH, HV or VH, VV） modes SAR data. Then the single-, dual-, full-polarization SAR data and new SAR characteristic information were used to analyze quantitatively the classification accuracy based on the Support Vector Machines （SVM）. The results show that classification overall accuracy of single-polarization SAR data is poor, and the highest is 56.53% of VV polarization. The classification overall accuracy of dual-polarization SAR is much better than single-polarization, the highest is 74.77% ofHV ＆ VV polarization data. The classification overall accuracy of full-polarization SAR is 80.21%, adding the difference characteristic information, ratio characteristic information and the first principal component （PC1） respectively, the overall accuracy increased by 3.09%, 3.38%, 4.14% respectively. When the full-polarization SAR data in combination with the all characteristic information, the classification overall accuracy reached to 91.01%. The full-polarization SAR data in combination with the band math characteristic information or the PC1 can greatly improve classification accuracy.

Separation of ocean features from multifrequency full-polarization SAR imagery

Different ocean features usually appear in synthetic aperture radar(SAR)images simultaneously.This makes the image complicated and hard to understand.Because of lower signal-to-noise rate,it is much more difficult to separate different ocean features than to separate different land features.A completely novel method is presented to separate ocean features from multifrequency polarimetric SAR imagery.AIRSAR data from Jet Propulsion Laboratory(JPL),National Aeronautics and Space Administration(NASA)are used in the case studies and good results are achieved.