中文摘要：

将双域滤波和三维块匹配算法相结合,充分利用了视频序列的时空频域的相关性,进行视频去噪。将三维块匹配视频去噪算法中小波阈值去噪得到的图像替换原引导图像,抑制了双域滤波算法基本层不平滑的问题。参考残差矩阵的能量谱对短时傅里叶变换系数进行阈值收缩,减少细节层的噪声残留。实验结果表明,本算法有效地解决了双域滤波算法的基本层不平滑、细节层有噪声残留的问题,主观和客观对比也表明,本算法有着较为优秀的去噪效果。

英文摘要：

The dual-domain filtering was used to remove the video noise combining with three dimensional block-matching. The correlation of time-space domain for the video sequence was fully considered. The noisy image was replaced by the result of wavelet thresholding in three dimensional block-matching video denoising algorithm as guide image, so that the base layer was smoother. Then, considering the energy spectrum of the residual matrix, the short-time Fourier transform coefficients were shrinkage which decreased the noise of the detail layer. Experimental results show that the proposed algorithm has better base layer and detail layer than the original dual-domain filtering. The subjective and objective comparison between different algorithms also proves that proposed algorithm has better denoising result.