A Fully Automated High-Performance Image Registration Workflow to Support Precision Geolocation for Imagery Collected by Airborne and Spaceborne Sensors

Abstract

Deriving precise coordinates from airborne and spaceborne imagery, with uncertainty estimates, is very challenging. Doing so is significantly more difficult when imagery is coming from one or more sensors that have questionable and/or incomplete photogrammetric metadata. Before precision geolocation activities can take place, that metadata must be complete and consistent such that images are correctly registered to one another and the Earth’s surface. This chapter describes an automated, high-performance image registration workflow that is being built at Oak Ridge National Laboratory to meet this need and focuses on the core concepts and software libraries underlying its creation. Highly encouraging initial system performance metrics are included as well.