Low complexity image coding technique for hyperspectral image sensors

Memory management and coding complexity are the major challenging issues of any hyperspectral image sensor. The hyperspectral image compression algorithm plays a greater role to improve the hyperspectral image sensor performance and save sensor memory. Many compression algorithms for hyperspectral images were proposed in past. The wavelet transform-based set partitioned hyperspectral image compression algorithms generate embedded output bit stream and also perform both lossy & lossless compression which makes them an ideal choice for any type of image sensor. The set portioned image compression algorithms use linked list or state table or markers to track the significance or insignificance of the block cube or coefficients. The linked lists grow with the bit rate which creates memory management issue while state tables or marker size is fixed which is not favorable with the low bit rate. In this study, a novel implementation of the set partitioned compression algorithm is proposed which employs parallel processing to reduce the coding complexity and exploits the linear indexing of the wavelet transform to track the set or coefficients to save the coding memory. The simulation results show the proposed compression algorithm 3D-BCP-ZM-SPECK reduces the coding complexity multiple folds with no need of coding memory.