Compression for hyperspectral images using three dimensional wavelet transform

In this paper, we apply the three dimensional wavelet transform to hyperspectral images. In particular, in order to compress hyperspectral data, we propose to use the three dimensional version of the set partitioning in hierarchical trees (SPIHT) algorithm, which has been successfully applied to 2 dimensional images and video signals. In order to evaluate the performance of the three dimensional SPIHT algorithm, we compute the SNR of compressed images and classification accuracies in the original images and the reconstructed images. Experiments with AVIRIS data show that high compression is possible with negligible information loss.