A.L. Priorov, V.V. Khryashchev, M.N. Golubev
The model of impulse noise with random values of impulses is considered. The comparative analysis of several modern filtration algorithms with classical rank-order statistics filters is carried out. In research were used both numerical restoration errors, and visual expert quality estimations. Recommendations for the best filtration algorithm for different classes of entrance images and impulse noise probability are resulted.