Adaptive modulated wavelet subband image coding

High quality, embedded, variable rate image compression can be achieved by decomposing an image into subbands of different sizes, modeling with one significant AM-FM component as well as encoding with a distinct procedure for each subband. In this paper, an adaptive subband decomposition using the modulated wavelet packet transform is proposed, where the adaptation of modulating frequencies is based on the energy spectral density in a resolution-recursive framework; and, each subband can be well represented by the modulated wavelets. An extension of SPIHT algorithm that is needed for encoding the complex-valued modulated wavelet coefficients is also presented. Experimental results show that the modulated wavelet subband image coding is preferable to both the wavelet coding and the JPEG standard.