A method of data compression for 3D spectrum-image data based on contour-texlure decomposition is described. According to the features of 3D spectrum-imagedata is produced by an imaging spectrometer. One of the 2D monochromatic imageswhich constitute the 3D spectrumimage data is decomposed into a set of contours andthe texture surounded by their contours with contour-texture decomposition. It is thefact that 3D spectrum-image data are a pile of monochromatic forages which extendalong spectral direction

one scene on ground yields the same contour in eachmonochromatic image With the exception of texture. It is enough to encodethe contours of one of the monochromatic images. For the texture variety

we encodethe polynomial coefficients corresponding each monochromatic image. The experimentwith the original 3D spectrum-image obtained by the 64 Bands Airborne ImagingSpectrometer developed by our country in the region Darwin

Australia shows that acompression ratio 379:1 can be achieved when the texture coefficients are expressed bya zero order polynomial.