Textures, game 'life' and non-linear dynamic systems

A new, algorithmic approach is proposed for analysis and synthesis of textured images. A notion of evolutionary models with feedback is introduced. As an example of a simple evolutionary model, a popular mathematical game, called Life, operating with binary images in a rectangular raster is considered. Evolutionary behavior of this model is studied analytically and experimentally for random binary images. Effects of deterministic and stochastic boundaries are discussed. Evolutionary behavior of the model as a function of its parameters is investigated. Phenomena of a 'stable chaos', 'chaos ordering', and 'chaos regeneration' are mentioned. In conclusion, the authors discuss the connection between the model and the deterministic chaos theory, autowave processes, and neural networks.