Generation of "similar" images from a given discrete image

A discrete image of several colors is viewed as a discrete random field obt ained by clipping or quantizing a Gaussian random field at several levels. Given a discrete image, parameters of the unobserved original Gaussian rand om field are estimated. Discrete images, statistically similar to the origi nal image, are then obtained by generating different realizations of the Ga ussian field and clipping them. To overcome the computational difficulties, the block Toeplitz covariance matrix of the Gaussian field is embedded int o a block circulant matrix which is diagonalized by the fast Fourier transf orm. The Gibbs sampler is used to apply the stochastic EM algorithm for the estimation of the field's parameters.