The following content has been provided by the University of Erlangen-Nürnberg.

Welcome everybody to the lecture Diagnostic Metric Image Processing.

So we start last week with MR intensity in homotenities.

So to repeat what we learned yesterday, you can see here at the bottom row,

you can see the observed image on the left side.

So you can see, okay, here is a little bit brighter

and at the bottom it's a little bit darker.

So this is the gain field in case

if we have a multiplicative model

or the bias field when we have an additive model.

And this is the ideal image or restored image

we want to work with.

So yesterday we learned about a method

called homomorphic unsharp masking.

We can see here the formula.

So g are the observed image intensities

divided by the bias field

and you get out the restored image.

And we did this with me and meij.

So maybe someone can explain, okay,

about the mean how or how to compute this method.

How does it work?

Nobody was here yesterday.

Okay, so maybe we start in front.

Maybe you can give an intuition how the method works.

Homomorphic unsharp masking.

You have the energy.

You compute the mean.

Sum up and divide the number of the sum.

And you have the bias field.

And you have to divide by this.

Okay, so it was not all, not all was correct.

So maybe we continue with the next lady.

So maybe you can say what was wrong.

Nope.

Okay, next one.

Okay, then maybe a guy.

You, also not.

Okay, so it was not clear yesterday what we did.

Ah, okay, good.

Next guy was here.

Okay, okay, okay.

You were here yesterday, so I know this.

So what we do, we calculate the mean

from the neighboring pixels.

So we have the new one here,

and then divide it by the bias field.

Okay, okay.

Okay, so the me was calculated, so this was correct.

You compute the mean of the whole image.