A. A. Sirota, E. V. Voronova
Computer modeling of area localized objects on the uniform background with neural net usage is considered in detail. This program’s algorithm contains the three parts: background texture generation with causal autoregression models usage, the object images modeling by neural nets with radial basis functions, and the object images superpositioning on the background with autoregression half causal and noncausal models usage. The main feature of this algorithm is a transparency mask usage to store object cutout and to remove a background from objects images.