Differential prediction-based lossless compression with very low-complexity for hyperspectral data

This paper presents Optimized Combination Coding (OCC) as a differential prediction-based (DP-based) lossless compression scheme for hyperspectral data with very low complexity. The main modules of OCC scheme consist of two predictors and Huffman codec. The pixels are processed differently according to their location in the image. The prediction errors between predicted and real pixel gray value are encoded ultimately into variable-length codewords attached to the binary code stream by Huffman codebook. Furthermore, the comparison between JPEG-LS algorithm and OCC scheme shows that optimized compression parameter of JPEG-LS algorithm slightly outperforms the scheme, however, the level of complexity described is fairly lower than the JPEG-LS ones. Additionally, several key issues of the codec implementation by means of FPGA technique are introduced.