A Polarization Decomposition Algorithm Use For Urban Building Targets With Large Azimuth

Man-made targets(e.g.,buildings)in urban areas are usually of regular geometry.The backscattering from the targets greatly changes as their azimuth angles vary with the reference to the incoming radar wave.When the azimuth angle is close to 45�,their backscattering often contains strong volume scattering.In some existing algorithms to decompose polarimetric signatures of a target,the volume scattering is usually used as the basis for discriminating vegetation areas from the non-vegetation target,but the double-bounced scattering as the basis for discriminating urban areas from the non-urban ones.Then,the backscattering characteristics from the urban areas with large azimuth and from vegetated areas are similar.Thus,misclassification occurs when these decomposition algorithms are used.To solve this misclassification problem,we proposed a new algorithm that is based on the subaperture technique.The subaperture technique is a technique to obtain two or more images with different azimuth angles from a single PolSAR image through the subaperture decomposition.In this study,two subaperture images with a range of azimuth angles are obtained following the selection principle for the aperture ratios.As the backscattering from urban targets varies with azimuth angles,the urban correction parameter is extracted from the subaperture images.The parameter can well characterize the backscattering from urban areas with a large azimuth angles and then is implemented into the existing decomposition algorithm.Thus,misclassification can be reduced.To verify the validity and effectiveness of the algorithm,we applied the algorithm to datasets of three spaceborne sensors,RADARSAT-2,Canada;GAOFEN-3,China;and ALOS2/PALSAR,Japan,and all achieved good classification results.Compared with the Pauli decomposition algorithm,this algorithm can reduce the volume scattering of urban areas with large azimuth angle by about 50% while ensuring that the classification results of other areas are not changed.Finally,the influence of the change of azimuth angle and aperture ratio on the classification results was articulated,and the relationship between the azimuth angle and urban correction parameter is quantified,and new algorithms are implemented to experiment with a variety of different aperture ratios,and the aperture ratio of the new algorithm is determined to be 0.75.