Reducing noise in high-resolution video sequences by using wavelet shrinkage in the temporal-spatial domain

A novel method for reducing video noise by using wavelet shrinkage in the temporal-spatial domain was devised and evaluated. In a temporal-spatial wavelet transform, static areas in a picture appear in a temporal low-frequency sub-band. In regards to this sub-band, the wavelet shrinkage can use a high threshold value in a shrinkage function and attains good noise-reduction performance. On the other hand, moving areas in a picture appear at the temporal high-frequency sub-band in a wavelet transform. In this sub-band, the wavelet shrinkage can not use a high threshold value, so a combined algorithm with a spatial-median filter is used. The noise-reduction performance of the proposed method was confirmed by a subjective evaluation, which showed that the quality of the noise-reduced images produced by the devised method is superior to that produced by traditional noise-reduction methods.