An automated approach for registration of multisensor/multi-resolution imagery

In this study, an automated method that is based on image intensities for geometric registration of multi-sensor/multi-resolution imagery acquired from EO-1 Hyperion and IKONOS satellite platforms is proposed. Method performs an area-based transformation to register images of different spectral and spatial resolution with high geometric accuracy. Method basically compares similarity of intensities of two images. Position where there is highest similarity score in translated blocks gives the best match. Chrominance Transform Operation (CTO) that we tested in this study as a similarity measure depicted higher accuracy and performance compared to methods commonly used in the field i.e. Normalized Cross-Correlation (NCC) and Mutual Information (MI).