New approach for noise cleaning on real time videos based low-pass filtering and mathimatical morphology

Video de-noising presents a necessary step in video processing. It improves its quality and permits a preview processing for other applications such as objects tracking, compression, feature extraction, edge detection, motion tracking... etc. In this paper, we combined the mathematical morphology operations with low-pass filtering to improve real-time video de-noising corrupted by different noise types. Different kinds of morphologies and filters are proposed in order to define the best combination possible. This algorithm does not require any motion estimation. It isn´t present a blur in de-noising video. Nevertheless, it can be a good reference for estimate motion in video.