Efficient compression of hyperspectral images by grouping around lines

In this paper we present a new lossless algorithm for compression of the signals from advanced hyperspectral infrared sensors onboard sun- and geo-synchronous environmental satellites. At each stage, our compression algorithm achieves an efficient grouping of channel data points around a relatively small number of I-dimensional lines in a large dimensional data space. The parametrization of these lines is very efficient, which leads to efficient descriptions of data points via adaptive clustering. Using one full day's worth (24 h) of global hyperspectral data obtained by the AQUA-EOS Atmospheric Infrared Sounder (AIRS), the mean ratio of compression attainable by the method is shown to be similar or equal to 3.7 to 1.