Differential Video Noise Estimation
Zhu Lei, XU Pei-xia · 2006
In this paper, we propose a fast and reliable white- noise variance estimation. The method subtracts two sequential frames of video first and then finds intensity-homogeneous blocks in both original image and differential image, and at last estimates the noise variance in these blocks by a Gaussian weighted averaging process. Experiments show that the proposed method performs well both in highly noisy and good- quality images. It also works well in videos including rich motion, large textured areas and few uniform blocks.