We present a new classification scheme, dubbed spectral classification
, which uses the spectral characteristics of the image blocks to class
ify them into one of a finite number of classes, A vector quantizer wi
th an appropriate distortion measure is designed to perform the classi
fication operation. The application of the proposed spectral classific
ation scheme is then demonstrated in the context of adaptive image cod
ing, It is shown that the spectral classifier outperforms gain-based c
lassifiers while requiring a lower computational complexity.