The problem of minimax estimation for the infinite-dimensional stochastic model is considered. The prior information about the random elements involved is formulated in term of second-order moment characteristics. The minimax estimation procedure is described and the corresponding numerical algorithm is presented. It is proved that the least favorable distribution of the model random elements is Gaussian. The efficiency of the proposed estimation algorithms is illustrated by means of the examples related to the signal and field robust processing.