Separation of Built-up Areas Using Polarization Orientation from Polarimetric Sar Images

Polarimetric decomposition and classification are important applications for polarimetric synthetic aperture radar (POLSAR) images. Among many methods developed so far, entropy-anisotropy-alpha classification [1] and three component decomposition [2] are most popular. In an urban area analysis, it is found that the polarimetric analysis has a problem to identify built-up areas. With the three component decomposition technique, built-up areas are assigned to a strong volume scattering target. In order to extract built-up areas, the method based on polarimetric correlation coefficient in both linear and circular polarizations was proposed and its performance was demonstrated using X-band airborne POLSAR data [3]. When this method is applied to L-band POLSAR data from ALOS PALSAR, extraction of built-up areas is insufficient. In this paper, polarization orientation angle shifts are investigated to separate built-up areas.