Doubly stochastic models of images

Problems of synthesis of doubly stochastic autoregressive models are considered that can be used for the formation of images close to real signals in their properties. A mechanism is proposed for the formation of images with parameters close to the given characteristics of real signals. The main feature is the formation of parameter fields by fitting to a given mathematical model of the image. In addition to the solution of the synthesis problem, the problems of filtration of multidimensional images generated by doubly stochastic models are considered. Various filtration algorithms are considered, in particular, nonlinear Kalman filtration, multistage Kalman filtration, and Wiener filtration.