论述了几种基于时域和空域混合的对低信噪比视频图像进行降噪处理的算法。考虑单独使用时域或空域降噪算法对视频图像进行处理的缺点和不足。在时域上，由于视频图像帧与帧之间有较强的相关性，采用递归加权算法；在空域上，分别采用第二代 Curvelet变换理论的 Wrap算法（Wrapping-Based Transform）、全变差（TV）算法、 Lee降噪算法、基于偏微分方程各向同性非线性扩散去噪算法（PM-AOS算法）以及中值滤波等降噪算法。实验结果表明时域和空域混合的降噪算法可提供较好的视觉效果，较好地保留视频图像的边缘、纹理等细节信息，超过单独使用时域或空域对视频图像降噪的算法，且部分算法处理时间相对较短，具有在硬件平台上移植的可行性。

This paper presents several mixture methods based on temporal domain and spatial domain of low SNR video denoising algorithms. The new methods take into account separately the insufficiency of temporal domain or spatial domain video denoising algorithm. In temporal domain, there is a strong link between video image frames and frames, so we adopt the recursive weighting algorithm. In spatial domain, we adopt Curvelet Transform theory of second generation algorithm (Wrap Wrapping - Based Transform), Total Variation(TV) algorithm, Lee’s image enhancement algorithm, the anisotropic nonlinear-diffused algorithm based on P-M equation (PM-AOS) and the median filtering algorithm, etc. Experiment results show that the effect of these temporal domain and spatial domain mixture denoising algorithms is obvious. These kinds of algorithms could provide better video quality, better retention of image edge and texture details. The processing time of several temporal domain and spatial domain mixture methods is shorter, and these methods have larger practical application on the processing of video denosing in the hardware platform through algorithm transplantation.