Onboard Lossless Compression of Hyperspectral Imagery Based on Hybrid Prediction

In this paper, we present an effective low-complexity algorithm for onboard lossless compression of hyperspectral images, the algorithm based on hybrid prediction. It is suitable to spacecraft onboard implementation as having much less complexity. Specifically, the proposed algorithm is mainly composed of three parts to compress the hyperspectral images. First of all, a three-dimensional (3-D) predictor, improved LCL-3D algorithm, is used to exploit the spatial correlation and spectral correlation efficiently, then a two-dimensional (2-D) nonlinear prediction algorithm is applied on the residual image after the 3D predictor. Finally, the residual image is entropy coded by the Rice coding. Performance of the method is compared to those algorithms, pure JPEG-LS, differential JPEG-LS, CALIC-extended, and LCL-3D. Simulation results show that the method outperforms LCL-3D as well as other compression algorithms and can be implemented onboard.