Digital filters play an important role in enhancing noisy images, some of nonlinear fillers are given, and a review of essential statistics and the types of noise are presented . The median filters are chosen for the present study. The following 11 hers are studied in terms of their statistical descriptions and mathematical formulation : i- Conventional median filter ii- Max/median filter. iii- Separable median filter. The effect of these median filters on different kind of noise and their effects on edges is demonstrated. A proposals for improved techniques is presented; namely, the iterated conventional median (ICM) , the iterated max/median (IMM) , the iterated separable median (ISM) . The effect of each of them on two test images , one corrupted by additive Gaussian noise and the other by additive impulsive noise , is studied in terms of: i- Noise removal, ii- Preservation of thin lines and sudden jumps in gray level, iii- Preservation of edges and comers . As a quantitative tests for the fillers performance ,the mean square error MSE and the local statistics (local mean and local variance) for both noisy and filtered images are calculated and compared . Linear and non - linear filters performances are also demonstrated in this study and compared quantitatively in terms of MSE and the local statistics . Some of the present fillers and the proposed filters are applied on a real images and compared qualitatively .