Abstract
Ground control points (GCPs) are essential spatial references for geometric calibration, image registration, and distortion correction of on-orbit remote sensing payloads. However, low-resolution (LR) remote sensing images (RSIs), usually derived from the high Earth orbit (HEO) cameras, lack texture and intensity details, making traditional GCP extraction methods ineffective. To address these issues, this paper proposes a novel GCP extraction method based on the integrated multi-curvature and spatial structure (IMCSS) features for LR RSI, which eliminates the dependence on grayscale information and differs from conventional methods in three aspects: (1) Adaptive multi curvature thresholds based on the Region of Support (ROS) are used to detect feature points in landmark contours, thereby eliminating trivial local details and preserving robust feature points. (2) A novel rotation invariance feature descriptor is constructed by integrating multi-order curvature with the spatial distribution of neighborhood contours. (3) The combination of an unconventional sparse matching strategy and prior geometric constraints achieves a competitive correct matching rate. The proposed IMCSS method is evaluated on FengYun-4B LR RSIs using the Global Self-consistent, Hierarchical, High-resolution Geography (GSHHG) database as the reference, a widely adopted coastline dataset for GCP extraction. To verify the generalization of IMCSS across different spectral bands, further experiments are conducted using high-resolution Landsat8 and SDGSAT 1 images as references. Experimental results demonstrate the effectiveness of IMCSS for GCP extraction in high-orbit LR RSIs, achieving a correct matching rate of 100% and a matching accuracy of 0.5 pixels.
Funding Information
  • National Natural Science Foundation of China (62205083)
  • Hangzhou Joint Fund of the Zhejiang Provincial Natural Science Foundation of China (LHZSZ25F010001)
  • Open Fund of State Key Laboratory of Infrared Physics (SITP-NLIST-YB-2024-10)
  • Research Funds of Hangzhou Institute for Advanced Study, UCAS
  • Strategic Priority Research Program
  • Chinese Academy of Sciences (XDA19010102)
  • Research Funds of Hangzhou Institute for Advanced Study

This publication has 40 references indexed in Scilit: