基于Surfacelet 变换的3D Context 模型视频去噪新方法

摘要

本文将3D Context 模型应用于Surfacelet 变换域,提出一种新的视频去噪方法. Surfacelet 变换(ST) 是一种新的3D 变换,具有多方向分解、各向异性和低冗余度等性质. 根据视频信号ST 域内系数和噪声分布的特征,将2D Context 模型拓展到3D ,按照能量分布将ST系数分成多个子块,每个子块有独立的能量和阈值估计. 实验结果表明,本文算法噪声抑制效果明显优于分层2D 去噪声方法和其它现有的3D 方法,去噪视频的PSNR 值提高了约2dB. 从视觉效果来看,本文算法在去除噪声的同时,能很好的保留视频图像细节,运动物体非常平滑,有效解决传统算法中存在的拖影、闪烁等问题,尤其适合于包含剧烈运动和丰富纹理图像的视频.Title: A Novel Video Denoising Method with 3D Context Model Based on Surfacelet Transform　Abstract : 　We propose a novel video denoising method with 3D Context Model in Surfacelet Transform Domain (3DCMST) in this paper. In order to take advantage of the characteristic of the coefficients ,the Context model was extended from 2D to 3D. The ST coefficients were divided into several parts according to their energy distribution by 3D Context model and each part had independent energy estimate and threshold. Experimental results show that the proposed method achieves better denoising performance than other 3D or hierarchical 2D denoising methods , and remarkably improves the PSNR of video about 2dB. In terms of visual quality ,the proposed method can effectively preserve the video detail ,and the trajectory of motion object is very smooth ,which is especially adequate to process the video frames with acute movement and plenty of texture.Key words : 　video denoising ;Surfacelet transform;3D Context model ; directional filter bank