[daip] [Fwd: Re: [Fwd: Re: IMEAN and noise]]

[Eric Greisen]

Another way to look at the correlation does not matter to estimating
the noise.  Consider an image but you make a histogram of only every
5'th pixel in each direction.  Those pixels are independent.  If the
noise is constant across the image then you should get the same
histogram (less well determined perhaps) as if you included all pixels
and hence get the same width or rms.

Eric Greisen