High-resolution optical satellite geospatial products play a pivotal role in Earth observation science, as their geopositioning accuracy significantly impacts various geospatial applications. While conventional geometric correction methods rely on high-accuracy ground control points (GCPs), the global scarcity of such reference data has driven the development of GCP-free methodologies. Existing approaches, utilizing public base maps [e.g., Google Earth, shuttle radar topography mission (SRTM)] to generate coarse-accuracy GCPs (CAGCPs), predominantly employ single-temporal references, inherently constrained by the base maps’ intrinsic positioning errors. Although recent attempts to manually extract multitemporal control points have partially reduced random planar errors, these methods critically neglect substantial systematic and gross errors in both elevation and planar dimensions, particularly from off-ground features in urban areas. To further improve geopositioning accuracy using multitemporal base maps, this article introduces a novel method of semantic-enhanced geopositioning refinement (SEGPR), which refines the available CAGCPs through the incorporation of semantic information and error adjustment techniques with multitemporal base maps. Specifically, this article employs a robust method to automatically capture coarse-accuracy matching control points (CAMCPs) using deep features from multitemporal base maps. Subsequently, this article innovatively explores the error characteristics of CAMCPs on both ground and off-ground features. Based on this exploration, semantic-guided random sample consensus (SGRANSAC) is introduced as a coarse correction to address elevation systematic errors, planar projection errors, and gross errors in off-ground MCPs, ensuring reliable MCP selection. In addition, the proposed elevation-augmented weight iteration (EAWI) iteratively reduces the influence of systematic elevation and random planar errors related to ground MCPs, enhancing GCP accuracy, particularly in elevation. The proposed algorithm is evaluated on geometric positioning and modeling accuracy. Experimental results across five global regions demonstrate an average geometric positioning accuracy of 1.42 m, a 55.90% improvement over state-of-the-art nonbasemap methods, and a 50.52% enhancement compared to state-of-the-art basemap methods. All corrected products achieve absolute geometric positioning accuracies better than 2 m, and modeling accuracy improves by 18.37%. The implementation of the proposed algorithm is available athttps://github.com/graduate-2024/SEGPR

This paper aims at achieving fine-grained building attribute segmentation in a cross-view scenario, i.e., using satellite and street-view image pairs. The main challenge lies in overcoming the significant perspective differences between street views and satellite views. In this work, we introduce SG-BEV, a novel approach for satellite-guided BEV fusion for cross-view semantic segmentation. To over-come the limitations of existing cross-view projection methods in capturing the complete building facade features, we innovatively incorporate Bird's Eye View (BEV) method to establish a spatially explicit mapping of street-view features. Moreover, we fully leverage the advantages of multiple perspectives by introducing a novel satellite-guided reprojection module, optimizing the uneven feature distribution issues associated with traditional BEV methods. Our method demonstrates significant improvements on four cross-view datasets collected from multiple cities, including New York, San Francisco, and Boston. On average across these datasets, our method achieves an increase in mIOU by 10.13% and 5.21% compared with the state-of-the-art satellite-based and cross-view methods. The code and datasets of this work will be released at https://github.com/yejy53/SG-BEV.

Feature matching determines the orientation accuracy for the High Spatial Resolution (HSR) optical satellite stereos, subsequently impacting several significant applications such as 3D reconstruction and change detection. However, the matching of off-track HSR optical satellite stereos often encounters challenging conditions including wide-baseline observation, significant radiometric differences, multi-temporal changes, varying spatial resolutions, inconsistent spectral resolution, and diverse sensors. In this study, we evaluate various advanced feature matching algorithms for HSR optical satellite stereos. Utilizing a specially constructed dataset from five satellites across six challenging scenarios, HSROSS Dataset, we conduct a comparative analysis of four algorithms: the traditional SIFT, and deep-learning based methods including SuperPoint + SuperGlue, SuperPoint + LightGlue, and LoFTR. Our findings highlight overall superior performance of SuperPoint + LightGlue in balancing robustness, accuracy, distribution, and efficiency, showcasing its potential in complex HSR optical satellite scenarios.