Geo-Positioning Accuracy Improvement of Multi-Mode GF-3 Satellite SAR Imagery Based on Error Sources Analysis

Area of Science:

• Remote Sensing
• Geomatics Engineering
• Photogrammetry

Background:

• The GaoFen-3 (GF-3) satellite is China's first C-band, full-polarization SAR satellite.
• Improving geometric performance of SAR images is crucial for accurate geospatial applications.

Purpose of the Study:

• To enhance the geometric performance of multi-mode GF-3 satellite SAR images.
• To develop error sources-based weight strategies for geometric improvement without Ground Control Points (GCPs).

Main Methods:

• Robust SAR image registration and SAR-features from accelerated segment test (SAR-FAST) for tie point extraction.
• Space intersection method for tie point object-space position calculation.
• Block adjustment with a bias-compensated rational function model (RFM) and DBSCAN clustering.
• Error sources analysis to develop weight strategies for normal equation matrix and Preconditioned Conjugate Gradient (PCG) method for solving.

Main Results:

• The proposed method successfully improves the geometric positioning accuracy of GF-3 satellite SAR images.
• Accuracy improvements were achieved within 2 pixels without the need for GCPs.

Conclusions:

• Error sources-based weight strategies effectively enhance the geometric performance of GF-3 SAR images.
• The developed approach offers a viable solution for accurate SAR image georeferencing.