Lightweight Remote Sensing Image Storage via GIS-Cloud Architecture with Fractal-DTCWT Compression and Knowledge Graph Indexing

Abstract

This paper proposes a lightweight and secure storage framework that integrates geographic information system (GIS) management, cloud computing elastic architecture, fractal dual tree complex wavelet transform (DTCWT) compression, and knowledge graph indexing to address the storage and retrieval challenges brought about by the surge of multi-source heterogeneous remote sensing image data. This method first removes redundancy through band correlation analysis at the GIS end, and uses DTCWT and fractal coding to achieve high fidelity compression; Subsequently, a distributed database was built in the cloud, and a unified semantic index was generated using a knowledge graph. Chaos sequence algorithm was introduced to encrypt the index. To verify performance, the experiment used 500 high-resolution satellite remote sensing images (in two sizes of 512 × 512 and 1024 × 1024) as the dataset. The results showed that the reconstructed image had a maximum information entropy of 9.94 (with an expected standard of 8), and the details were preserved intact; The compression ratio remains stable above 6.0 (expected standard 5.5), and there is no significant decrease in land use classification accuracy when the compression ratio is 6.0; The MSSIM index based on knowledge graph is higher than 0.947, indicating high retrieval efficiency; The pixel change rate of chaotic encryption can reach up to 99.4%, with strong security. This study provides a practical and feasible technical solution for efficient storage, secure sharing, and fast retrieval of massive remote sensing data