Noise estimation in an image

2012 1 11
1 个回答
4 次查看（30 天）

0 个投票

Hello,
When I want to estimate the noise variance of an image, I use to select an homogeneous region and compute the variance in that area. But I found another way to compute it in the web that I don't really understand and which yields the same results :
function [sigma,eps1]=estimateNoise(im)
h=zeros(3,3); h(2,2)=1;
h(2,1)=-1/4; h(2,3)=-1/4;
h(1,2)=-1/4; h(1,3)=-1/4;
f=sqrt(4/5);
eps1=f*imfilter(im,h,'replicate');
sigma=sqrt(1/numel(im)*sum(eps1(:).^2));
Can someone explain why is it the same ?
Regards

请先登录，再回答此问题。

回答（1 个）

scaramanga
scaramanga 2012-1-11

0 个投票

By the way, estimateNoise yields the standard deviation (and not the variance).

类别

在 帮助中心 和 File Exchange 中查找有关 Image Category Classification 的更多信息

标签

提问:

scaramanga
scaramanga
2012-1-11

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by