我发现从降低图像噪声的一些方法，但我的任务就是来衡量它。

I've found a few ways of reducing noise from image, but my task is to measure it.

所以，我感兴趣的算法，会给我一些数字，噪音等级。与这个数字我可以说是一个图像具有噪声比别人少。

So I am interested in algorithm that will give me some number, noise rating. That with that number I will be able to say that one image has less noise than others.

推荐答案

要解决我的问题，我用另一个方法：

To solve my problem I used next approach:

我的噪音等级仅仅是数字，被认定为噪声像素。要区分正常的噪音像素，我刚才计算其相邻像素的中间值，如果它的值大于某个临界值，我们说这个人是噪音。

My noise rating is just number of pixels that were recognized as noise. To differentiate normal pixels from noise, I just calculated the medium value of its neighbor pixels and if its value was bigger than some critical value, we say that this one is noise.

if (ABS(1 - (currentPixel.R+currentPixel.G+currentPixel.B)/(neigborsMediumValues.R + neigboursMediumValues.G + neigboursMediumValues.B))) > criticalValue)
then
{
    currentPixelIsNoise = TRUE;
}