Comparison of diverse approaches for synthetic aperture radar images pixel fusion under different precision registration

All kinds of approaches have been proposed to implement the pixel fusion for remote-sensing images that must be strictly registered in advance. Errors introduced by finite precision registration would influence the performance of fusion process. Moreover, it is not easy to achieve high precision registration on synthetic aperture radar (SAR) images. Thus it is significative to evaluate the performance of diverse fusion approaches under different precision registration, especially in low precision condition. In this study, several current image fusion methods are applied to execute the pixel fusion on real SAR images under different precision registration, and the influence of registration error on their respective performance is tested. The experimental outcomes indicate that the contrast pyramid and wavelet transform fusion methods have shown the better performance than the other compared methods.