Video Denoising Based on Spatial-Temporal Filtering

Video noise reduction based on temporal spatial recursive filter isproposed in this paper. In the proposed model the recursive time-weighted average is applied to the areas where the motion has notbeen detected. The new model is able to be adaptive in each areadepending on whether the area is static or movable. More precisely, more noise removal will be done in the static areas, and lessremoval in the motion areas. In the new model, noise reduction iscarried out in three stages. At first, noise is filtered withspatial filter. Secondly the temporal filter is applied for morenoise repression. Thirdly motion detector and recursive timeaveraging are applied to improve the temporal filter's output. Thequantitative and qualitative performance in various noise levelsdemonstrates the superiority of the proposed scheme as compared withsome well-known methods.