Lossless and gradual coding of hyperspectral images by lifting scheme

In this paper, we are interested in compactly coding hyperspectral images for applications requiring a progressive and exact reconstruction. For this purpose, a two-step procedure is envisaged. Firstly, a spectral clustering of the whole data set is carried out. Then, within each spectral cluster, a band-ordering of the images is operated and the spectral and spatial similarities are simultaneously taken into account through an extended version of a lifting-based decomposition scheme. Our contribution relies on coupling these two steps and in optimizing the band-ordering procedure.