SCORE: an algorithm for stochastic compression and regeneration of images
Nesterov A. Yu.
"Математика. Компьютер. Образование". Cб. трудов XI международной конференции. Под общей редакцией Г.Ю. Ризниченко Ижевск: Научно-издательский центр "Регулярная и хаотическая динамика", 2004. Vol. 2. Pp. 555-564.
Appreciable reducing the size of a graphic file (in comparison with the original raster represen-tation as a matrix of coefficients) seems feasible since partial loss of raw data is not fatal, thus every compression technique makes a compromise between the compression ratio and fidelity of regenerated images. It is quite clear theoretically that the possibility of data compression itself springs from data redundancy, and in the case under review that is proximity of brightness levels in neighboring pixels, however, corresponding mathematical constructions in the spirit of random processes are ignored by existent algorithms. The proposed method SCORE considering an image as a realification of a two-dimensional stochastic field takes its statistical properties (say, bivariate correlation function) into account and, consequently, embodies the optimum (in the sense of mean square error) handling.
|
