A Survey on Hyperspectral Remote Sensing Image Compression
Fan Zhang, Chao Chen, Yuting Wan · 2023
Hyperspectral remote sensing images (HSI-RS) capture the fine spectral information of terrain, but also bring large data volume, which poses a huge challenge to the storage, transmission and even processing. HSI-RS compression achieves image critical information representation lossless or lossy with high fidelity through effective spatial-spectral redundancy removal. As an extension of the research field of traditional natural image compression technology, there has not yet been a study concluding and comparing the various HSI-RS compression methods. The paper first introduces and explains the differences between the HSI-RS compression and natural image compression. Then the methods for HSI-RS compression are investigated and can be divided into five main categories: 1) transformed-based; 2) prediction-based; 3) dictionary-based; 4) decomposition-based; 5) learning-based. Experiments are carried out to make a comparison on the fidelity and rate distortion performance. At the end of the paper, the future development directions for HSI-RS compression are discussed.