Self-similarity of the multiresolutional image/video decomposition: smart expansion as compression of still and moving pictures

摘要

The paper introduces a new combined fractal/multiresolutional image compression based on the observed property of self-similarity of the pyramidal image transform. The gist of the method is zooming out from a (possibly shrunken) low-resolution image producing a sharp and crisp "natural looking" high-resolution view, without blockiness and jaggedness. It is demonstrated that the technique possesses features of preserving thinness of lines on expansion, translational invariance and providing a perfect high-resolution representation of the gradient fill. The multiresolutional transform algorithms and 'smart' image magnification developed for still images have been generalized to deal with moving pictures as a three-dimensional, spatio-temporal frame sequence, which permits rapid compression, smooth motion interpolation, and has potential for use in video transmission in real time.