Sensor-agnostic photogrammetric image registration with applications to population modeling
- ORNL
Photogrammetric registration of airborne and spaceborne imagery is a crucial prerequisite to many data fusion tasks. While embedded sensor models provide a rough geolocation estimate, these metadata may be incomplete or imprecise. Manual solutions are appropriate for small-scale projects, but for rapid streams of cross-modal, multi-sensor, multi-temporal imagery with varying metadata standards, an automated approach is required. We present a high-performance image registration workflow to address this need. This paper outlines the core development concepts and demonstrates its utility with respect to the 2016 data fusion contest imagery. In particular, Iris ultra-HD video is georeferenced to the Earth surface via registration to DEIMOS-2 imagery, which serves as a trusted control source. Geolocation provides opportunity to augment the video with spatial context, stereo-derived disparity, spectral sensitivity, change detection, and numerous ancillary geospatial layers. We conclude by leveraging these derivative data layers towards one such fusion application: population distribution modeling.
- Research Organization:
- Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)
- Sponsoring Organization:
- ORNL work for others
- DOE Contract Number:
- AC05-00OR22725
- OSTI ID:
- 1302901
- Country of Publication:
- United States
- Language:
- English
Similar Records
Stereo Camera A Auxiliary Data from Stereo-Photogrammetric 3D Cloud Mask System
Automated Image Registration (AIR) of MTI Imagery