Optical Data Assisted Ground Control Points Extraction Method from Gaofen-3 SAR Images

Ground control points (GCPs) are important basic data in surveying and mapping production, but the current field measurement methods cannot efficiently meet the needs of large-scale GCP collection. This paper proposes a GCPs extraction method for Gaofen-3 (GF-3) Ultra Fine Strip synthetic aperture radar (SAR) images with optical data assistance. First, we refine several calibration parameters of GF-3 SAR images using the Range Doppler model and a small number of known GCPs to improve the accuracy of image localization. Using lamppoles as the target GCPs, we design a lamppole target search method for spaceborne SAR images based on optical shadow features. The coordinates of lamppole pixels in SAR images are determined by enhancing narrow dark information and template matching. A SAR stereo positioning method is adopted to realize accurate positioning of the three-dimensional coordinates of the lamppole targets. The experimental results for Zhengzhou reveal that the mean error of the extracted GCPs is 2.40 m in plane and 1.58 m in elevation, which are feasible.