Enhanced satellite images block adjustment via preliminary RPCs correction and adaptive two-step tie points matching strategy

With the increasing diversification and complexity of satellite imagery, traditional uncontrolled block adjustment algorithms often encounter challenges such as significant disparities in image positioning accuracy and difficulties in weight assignment. This paper proposes utilizing an adaptive LNIFT algorithm to extract ground control point data from public reference Digital Orthophoto Maps (DOM) for preliminary RPCs correction. This can enhance the internal consistency of experimental data, making it easier to determine weights during the adjustment process. Furthermore, we introduce a tie point extraction strategy that integrates coastal and cloud mask data and incorporates a two-step matching strategy. This optimizes extraction effectiveness and ensures robust tie point extraction capability under complex scenes. Finally, block adjustment is conducted to further reduce internal geometric consistency. The experimental results indicate that this method can effectively improve the internal geometric consistency between images, and the proposed matching strategy significantly enhances the adjustment results.