A framework for SAR-optical stereogrammetry over urban areas

Area of Science:

• Geosciences
• Remote Sensing
• Photogrammetry

Background:

• Diverse earth observation data from remote sensing satellites require fusion techniques for integrated analysis.
• Synthetic aperture radar (SAR) and optical imagery fusion enables 3D information generation via stereogrammetric methods.

Purpose of the Study:

• Investigate the application of stereogrammetry pipelines to very-high-resolution (VHR) SAR-optical image pairs.
• Assess the potential of semi-global matching and epipolarity constraints for VHR SAR-optical data.
• Improve absolute geolocation accuracy of VHR optical imagery relative to VHR SAR imagery.

Main Methods:

• Application of semi-global matching for image matching in a multi-sensor setting.
• Investigation of epipolarity constraints to optimize image matching.
• Multi-sensor block adjustment using rational polynomial coefficients for geolocation refinement.

Main Results:

• Demonstrated applicability of semi-global matching for VHR SAR-optical pairs.
• Established epipolarity constraints to enhance matching accuracy and speed.
• Achieved improved absolute geolocation accuracy through multi-sensor block adjustment.
• Generated point clouds with a median accuracy of approximately 2 meters.

Conclusions:

• Confirms the potential of 3D reconstruction from VHR SAR-optical image pairs.
• Highlights the effectiveness of the proposed stereogrammetry pipeline for urban areas.
• Suggests significant advancements in multi-sensor data fusion for geospatial applications.