Block-Sparse Tensor Based Spatial-Spectral Joint Compression of Hyperspectral Images

摘要

A hyperspectral image is represented as a three-dimensional tensor in this paper to realize the spatial-spectral joint compression. This avoids destroying the feature structure, as in the 2D compression model, the compression operation of the spatial and spectral information is separate. Dictionary learning algorithm is adopted to train three dictionaries on each mode and these dictionaries are applied to build the block-sparse model of hyperspectral image. Then, based on the Tucker Decomposition, the spatial and spectral information of the hyperspectral image is compressed simultaneously. Finally, the structural tensor reconstruction algorithm is utilized to recover the hyperspectral image and it significantly reduce the computational complexity in the block-sparse structure. The experimental results demonstrate that the proposed method is superior to other 3D compression models in terms of accuracy and efficiency.