In this work, we address the problem of multichannel image retrieval in the compressed domain. A wavelet
transform is applied to each component of the multispectral image. The salient features are computed from
the resulting wavelet subbands. To this purpose, two approaches are envisaged. In the first one, the wavelet
coeffcients of each component are separately considered whereas in the second one, they are jointly processed.
More precisely, the contribution of this work lies on the fact that the features are extracted from the multivariate
distribution of the wavelet coeffcients modelized thanks to copulas. Experimental results indicate that the
second approach gives the best performances in terms of precision and recall.