Research On Video Denoising Algorithms Based On 3D Wavelets

Compared with traditional information format such as audio and character, image and video contain more information and are more intuitionistic and accurate. Hence,people pay much more attention on digital video.Research on digital video processing is meaningful both in theoretic study and practical application.However, video signal in reality is noised.It is rather important to extract valuable signal component before further analysis.Recently,multi-resolution and multi-direction analysis on multidimensional signal is a hot topic in academe.It is applied in many aspects of image and video processing.As for de-noising application,Donoho and Johnstone proposed Waveshrink in 1992,which is very successful in audio and image de-noising.Due to the design limitation of filter banks,expanding multi-direction analysis to 3-Dimensional signal never had a perfect solution until 2005,Yue Lu and M.N.Do overcome the difficulty with a real 3-D wavelets—Surfacelet.In this paper,we introduce a new multidimensional wavelets named Surfacelet in the beginning.Then,we review the traditional de-noising technique based on wavelets and emphasize on some key issues such as how to define a threshold value and threshold function.Meanwhile,we introduce another two 3-Dimensional wavelets transform—Separable Discrete Wavelets and Dual-Tree Complex Wavelets. We examine their theory in detail and compare their performance with Surfacelet in 3D data processing.The experiment results show that Surfacelet outperforms the others in video processing field.Finally,a wavelets threshold determine algo??m, considering the coefficients dependence of inter-scale and intra-scale,is proposed and applied to video denoising based on Surfacelet transform.The experimental results show that this de-noising algorithm can gain better PSNR as well as more decent visual effect.