An information extraction method for satellite remote sensing data
By constructing a three-dimensional description mapping table and a surface structure feature index table, and combining the graph convolution propagation mechanism, the problem of alignment and consistency modeling among data in satellite remote sensing data was solved, realizing collaborative analysis and high-order information extraction of multi-source remote sensing images, and improving information organization and cross-task adaptability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TIANXIELI (SHANDONG) SATELLITE TECH CO LTD
- Filing Date
- 2026-04-09
- Publication Date
- 2026-07-17
AI Technical Summary
Existing satellite remote sensing data information extraction technologies fail to effectively utilize metadata such as orbital parameters, imaging time and spatial coverage boundaries, resulting in insufficient perception of potential imaging logic and overlap relationships between data. This makes it difficult to achieve accurate content alignment and consistency modeling, limiting the collaborative analysis and high-order information mining of multi-source remote sensing images.
By constructing a three-dimensional description mapping table and a surface structure feature index table, the correlation consistency between images is analyzed, a set of related images is generated, and feature extraction is performed through a graph convolution propagation mechanism. Cross-image segment sequences are established, feature graph structures are constructed for correlation propagation analysis, and finally, information extraction results containing temporal context, spatial adjacency and surface structure correlation are generated.
It realizes joint encoding and aligned representation of multi-source remote sensing images, improves information organization and cross-task adaptability, enhances the consistency and traceability of results, and solves the problems of isolated local features and broken contextual relationships caused by heterogeneous and unevenly distributed data sources.
Smart Images
Figure CN122412652A_ABST