Multiple description based distributive compression for hyperspectral images

In our paper, we propose a distributed source coding scheme exploiting the principle of multiple description coding for a simple encoder implementation of hyperspectral image compression. Multiple descriptions hold great significance in scenarios where highly fidelity reconstruction is desired after transmission over error and loss prone transmission channels. Lossy Wyner Ziv coding is deployed in conjunction with multiple descriptions resulting in generation of multiple correlated independent substreams of key bands which are employed as side information at the decoder. In addition, adaptive parity generation is supported that provides a more dynamic reconstruction in terms of variable bit rate generation. The inherently compliant multiple description based distributive source coding paradigm is not only appropriate for limited onboard processing but also aids the establishment of independent processing nodes distributing the aggregate onboard computational load over multiple nodes for efficient transmission. The proposed scheme offers a thoughtful perspective on prevailing challenges in the design of robust hyperspectral imaging algorithms. Experimental results of PSNR performance depict that our scheme offers competitive performance as compared to various schemes in the same arena.