Image Registration Accuracy Analysis for Multi-Sensor and Multi-Temporal Remote Sensing Images

Image registration is the process of overlaying two images of the same sight taken at different times, from different view angles or from dissimilar sensors. Image registration was used in the first step for analysis of images acquired from one or more sensors having different spatial and / or temporal variations. Its main aim was to align two images geometrically. This study proposes automatic image registration method using the Image Auto sync workstation module was embedded in the ERDAS IMAGINE 16 software. This module generates hundreds of tie points between two images automatically to align more closely with the reference image. The registration accuracies were verified for the satellite images obtained from different bands, or different spatial resolution, and different platforms. Three image registration experiments were discussed in this paper to analyze the accuracy differences which have different spatial, temporal and radiometric resolutions. The results indicate that accuracy was high between same sensor and same resolution images, and low for different spectral and spatial resolution images.