Joint Source Channel Coding for Hyperspectral Imagery

On-board compression of hyperspectral imagery plays a vital role in the arena of remote sensing applications. This paper proposes a methodology to concatenate source coding with channel coding referred to as Joint Source Channel coding (JSCC) for hyperspectral images. The leverage of JSCC is that firstly the three on-board constraints namely memory, power and bandwidth are tackled with source coding. Secondly Channel coding mitigates the adverse effects of channel during the course of transmission of the downlinked image, thus maintaining the quality of the reconstructed image at the decoding side. Source coding exploits spatial and spectral decorrelation utilizing Intraband and Interband predictive coders. Serial Concatenated coders using Low Density Parity Check / Reed-Solomon (LDPC/RS) coder along with Convolutional coder are suggested for Channel coding. The proposed algorithm is simulated for Additive White Gaussian Noise (AWGN), Binary Symmetric Channel (BSC) and Rician channels. The methodology outcomes with an average BER of around 10−5 for different channel parameters, with an acceptable trade-off between complexity and performance.