A computational efficient algorithm for adaptive image compression

Presents an adaptive compression scheme for still images. According to the spatial activity of each block of the image, a different number of DCT coefficients are calculated. A lattice vector quantizer, where pyramids with different radius and dimensionality are used for the vector quantization of the coefficients of each block, is described. The proposed algorithm has a low computational complexity and results in high quality image reconstruction at low bit-rates per pixel.<>