| Image compression of hyperspectral is necessary because of the large storage requirements. The main objective of this research is to develop and implement fast and good predictors, for use in lossless compression of hyperspectral images. The algorithms developed in this research are compared in terms of compression performance and computational complexity against the best existent algorithms. These benchmark algorithms are LOCO-I and CALIC-Extended. Lossless compression algorithms are typically divided into two stages, a prediction stage to eliminate redundancy and a coding stage. The predictors can utilize pixels of the same spectral band, adjacent bands or both in order to perform the prediction. Algorithms found in documented research use information of their same band or an adjacent band, but not at the same time. An example of this type of algorithm is algorithm LOCO-1, which only uses the information in the same band in order to predict. However, the CALIC-Extended algorithm alternates between the information in the same band and that of the adjacent one. The algorithms developed during this research use pixels of both bands in order to do the prediction. The best lossless compression predictor implemented in this research, gave better compression than the state of the art along with a low computational complexity. |
