Co-Registration of Aerial, Terrestrial and Satellite Imagery

Co-registration of multi model geospatial images is essential for creating precise and integrated 3D spatial models. This paper presents an end-to-end method for Co-registration of UAV, Terrestrial and Satellite images to address difficulties due to varying perspectives and resolutions. UAV images were reconstructed in Reality Capture with the help of SIFT based feature matching. Terrestrial data collected from smartphone videography, extracted to frames and processed with COLMAP to enable sequential alignment due to vertical surface constraints. A hybrid workflow between Reality Capture and Cloud-Compare enhanced dense reconstruction accuracy, providing an RMSE of 0.403m by using iterative closest point refinement for the UAV and Terrestrial model. Satellite images were aligned via projection and resampling using ERDAS IMAGINE, enabling fusion across resolutions and finally the accuracy of the UAV, Terrestrial and Satellite Image model gives a measurement accuracy of about 1.2 m. The method enables scalable and precise geospatial integration, making it suitable for use in photogrammetry, mapping and remote sensing.