A new scheme of video denoising based on 3D wavelet transform

Video denoising using spatio-temporal information is considered to be more efficient compared with the use of just spatio information. In this paper, a new scheme for video denoising based on 3D wavelet transform is proposed. The input sequence is firstly decomposed into several high frequency frames and one low frequency frame by temporal decomposition. Then, an efficient adaptive threshold is introduced for the low frequency frame denoising while a simple universal threshold for the high frequency ones denosing. Experimental results show that the proposed scheme removes noise more significantly than those algorithms like SureShrink, Winener Filteering and MMSE estimation, which are used in 21) wavelet domain for video denoising without 3D-WT.