Image data compression by adaptive vector quantization of classified wavelet coefficients

The paper deals with a novel technique for image data compression based on adaptive vector quantization of wavelet coefficients. High compression factors can be achieved without loosing the informative content of each subband for the statistical properties of these coefficients. Performance of this approach, evaluated on a wide variety of pictures, is far better than the JPEG one, both in quantitative terms (PSNR) and in qualitative ones (absence of blockness distortion).