Analysis And Application Of Target’s Scattering Property In Polarimetric SAR Images

Polarimetric SAR is the combination of polarization, synthetic apenture, and radartechnologies, and is an important means of remote sensing. It contains a lot ofinformation which is vital in PolSAR image interpretation. This dissertation focuses onanalyses of scattering property of a target, and the innovations are summarized asfollows.1) The polarimetric property of a target is described by means of nearest coherenttarget extraction. The norm of a Kennaugh matrix is defined based on polarimetricsynthesis theory. A Kennaugh matrix is decomposed into two parts. One iscorresponding to a coherent target. The other is the residual matrix with minimum norm.The properties of the extracted coherent target are discussed from the point view of thespan and polarimetric characteristics.2) The polarimetric property of a target is described by deorientation operation.The orientation angle is an important parameter of a target in polarimetric radar remotesensing. There are many different definitions of the orientation angle for differentapplications. Eight kinds of definitions are compared and the relationships are analyzed.The properties of deorientation process are also discussed. The impacts of thedeorientation are analyzed. The double-bounce scattering coefficients before and afterthe deorientation operation are compared. These two coefficients are used forpolarimetric SAR image classification. Results show that the latter can distinguish threedifferent scattering mechanisms better than the former.3) The polarimetric property of a target is described by similarity parameter (SP).In the application of ship detection, SPs between the target and three canonicalscattering targets are calculated. The different scattering mechanisms between the targetand clutter are discussed in terms of SPs. The polarimetric cross entropy (PCE) isredefined based on the difference. Containing the value of span, a parameter namedspan polarimetric cross entropy (SPCE) is defined. SPCE is used for CFAR shipdetection and shows great efficience.4) The scattering property of a target is described by multifrequency andpolarimetric joint methods. The dependencies of coherency matrix between different bands are tested and the principal scattering component is evaluated by eigenvectordecomposition. The correlations between different bands are then estimated and thecoherency matrices in different bands are mapped onto a new larger matrix. The newmatrix can be used to describe the polarimetric property in multifrequency SAR image.A boxcar filter is then performed on the new matrix. The filtered data are put into acomplex Wishart classifier. Experiment result demonstrates that the proposed method issimple and has high classification accuracy.