中文摘要：

在对视频图像的获取过程中不可避免地会引入噪声，导致视觉效果下降。提出一种新的视频去噪算法。第一步时域滤波采用改进的自适应十字算法进行帧间运动估计，对于判别为没有运动的：区域莲用标准的加权均值时域滤波方法，对于运动区域，则沿运动轨迹进行滤波。第二步空域滤波借鉴小波分析框架和著名的非局部均值NLM（Non Local Mean）去噪算法㈦，对视频的每一帧进行空域分频处理。实验结果的分析与对比表明所提出的方法能有效地避免了运动模糊，较好地克服了平坦区域产生虚假纹理信息的问题，更好地保护了图像的边缘等细节信息。

英文摘要：

Noises are inevitably imported during the acquisition of video images and this leads to the degradation in visual effect. The paper proposes a new video denoising algorithm to address this problem. The first step is time-domain filtering, which adopts the improved adaptive cross algorithm to carry out inter-frame estimation. For the area where no motion is detected, the standard weighting average time- domain filtering is applied. For the area with motion, the filtering is conducted along the motion trajectory. In second step the spatial filtering, based on the framework of wavelet analysis and the famous non-local means denoising algorithm, every frame of the video is undergone spatial frequency division processing. Analysis and comparison of the experimental results demonstrate that the proposed denoising scheme can effectively avoid motion blurring, well overcomes the problem of false texture information produced in the fiat area, and better protects the specific information like the margin of the image.