A Robust SAR Terrain Self-Calibration Method Utilizing Projected Area Ratio

Abstract

The rapid advancement of synthetic aperture radar (SAR) satellites has enabled enhanced capabilities for all-weather and all-time Earth observation. Nevertheless, the geolocation accuracy of SAR images remains constrained by multiple error sources. Although ground calibration fields can substantially improve the geometric geolocation accuracy of SAR images, such procedures are typically time-consuming and labor-intensive. To address this issue, this article proposes an SAR terrain self-calibration method based on the projected area ratio (PAR), aiming to improve the geometric positioning accuracy of SAR ortho-images cost-effectively and efficiently. First, a novel concept, referred to as the projected area ratio, is introduced to quantitatively assess and compare the geometric distortions among different SAR images. This metric has a clear physical interpretation and enables effective discrimination between steep and flat terrain within SAR imaging regions. Subsequently, we propose a constrained generalized search-based clustering algorithm to efficiently extract tie points. The high correspondence between PAR maps and SAR images is leveraged to estimate SAR geometric error compensation. Finally, the proposed method is validated using four SAR images acquired over different terrain conditions, demonstrating that PAR can serve as a quantitative indicator of SAR geometric distortion. The applicability of the proposed method is further discussed based on the physical interpretation of PAR. Compared with the conventional GAMMA method, the proposed method demonstrates significant advantages in the matching success rate. In addition, corner reflectors are employed to quantitatively evaluate the geolocation accuracy of the corrected SAR orthorectified images. The experimental results show that the positioning accuracy is improved from approximately 80 to 3.952 m after applying the proposed method, demonstrating its effectiveness in enhancing SAR geolocation accuracy and enabling high-precision positioning for low-accuracy SAR satellites.