On a New Ship Detection Parameter Using Multi-polarization SAR Data

Abstract

In the present article, a new parameter is proposed to improve the accuracy of shipdetection based on polarimetric analyses of synthetic aperture radar (SAR) data. The parameteris similar to the normalized di®erence vegetation index (NDVI) for multispectral optical data,and it is de&macr ned as the di®erence of values in polarimetric entropy from coherence divided bythe sum of these two values. It has been known that the polarimetric entropy is high for targetsthat give rise to random scattering consists of di®erent scattering processes such as ships whileit is low for targets yielding single scattering, for example, surface scattering from sea surface.It is also known that the coherence, i.e., correlation between HH- and V V -polarization imagesof targets with volume or multiple scattering is low in comparison with targets to give rise tosurface scattering.DVI) for multispectral optical data,and it is de&macr ned as the di®erence of values in polarimetric entropy from coherence divided bythe sum of these two values. It has been known that the polarimetric entropy is high for targetsthat give rise to random scattering consists of di®erent scattering processes such as ships whileit is low for targets yielding single scattering, for example, surface scattering from sea surface.It is also known that the coherence, i.e., correlation between HH- and V V -polarization imagesof targets with volume or multiple scattering is low in comparison with targets to give rise tosurface scattering.