VERFAHREN ZUR KOMPRESSION VON BILDDATEN

摘要

Images are approximated by invariant densities of stochastic processes, for example by so-called fractals. The parameters of such stochastic process whose invariant densities optimally approximate the images to be compressed, according to a quality criterion, produce a compressed reproduction of the image. The optimal stochastic processes are found by means of a stochastic optimization process, preferably by means of a genetic optimization process that uses, among others, mutation, cross-over and exchange of genes in order to change genetic information. This process allows extraordinarily high data compression factors to be obtained, typically between 1000 and 10000. This process is suitable for data compression of still images as well as for the compression of images formed of moving image sequences.