In this paper we present the results of the impulsive noise filtering using the RM-KNN (Rank M-type K-nearest neighbor) algorithms justified by means of objective and subjective criterions of performance, the implementation in real time of the RM-KNN type filters for fixed images and video sequences, as well as their comparison with some filters already known by their good performance and investigated in different articles. The proposed filters reduce the noise of impulsive type and at the same time they preserve the fine details of the images for different applications like video conferencing, remote sensing and computer vision. The data processing was realized in real time by the use of DSP TMS320C6711.