A fuzzy video pre-filtering method for impulse noise reduction

A fuzzy video pre-filtering method (FVPFM), for reducing impulse noise and its mixture with other types of noise, consists of two stages: fuzzy detection and fuzzy filtering. The first stage shows a new high hit-rate method, which keeps nearly perfect coverage and correct rate of detecting noise. The high percentage is dependent on the accurate estimation of noise models. The second stage performs fuzzy smoothing by weighing the contributions of the inter-frame similarity or the intra-frame gradient. The new non-linear filter is specially designed for reducing video noise while preserving details and texture of successive frames. Experimental results are obtained to indicate real-time operation, pleasant PSNR and splendid video restoration by the proposed approach other than normal spatial methods.