Geolocation error compensation method for geocoded SAR images using pixel-offset series without control points

Abstract

SAR spatial geolocation of synthetic aperture radar (SAR) is crucial for SAR/interferometric (In)SAR applications. Geolocation errors often arise owing to factors such as platform instability, instrument degradation, and environmental variations. Current compensation methods rely on calibration field data or control points that are challenging to obtain on a large scale. An easy-to-use compensation method without control points was proposed through leveling-inspired adjustment using a pixel-offset series. The offsets between the SAR images were transformed into the offsets relative to a reference image with high-precision geolocation. Pixel offsets were averaged and treated as height differences along a leveling route, followed by leveling adjustments in both longitude and latitude. It is a product-level method that excludes imaging and interferometric parameters. Experiments with geocoded TanDEM-X SAR images near Guangzhou, China, demonstrate that the method (1) is independent of reference image position and can resist the gross geolocation error in geocoding to a certain extent, and (2) improves the root-mean-square error of the geolocation error at the tie-points to 0.72 pixels, a 65% reduction. The planimetric geolocation accuracy of the SAR image was better than one pixel, improved by 65% compared with Google satellite maps. These results demonstrate the applicability of the proposed method.

Keywords

• Synthetic aperture radar (SAR)
• Geocoding
• Geolocation error
• Planar block adjustment
• Leveling adjustment
• Pixel-offset series