Does it work?

Ah yeah, okay, now it works.

Good.

Let's start with a small quiz again.

So, most of you probably have it already.

Maybe you can make it live, so it is VOTE and then the code this time is BQHJ.

So, BQHJ.

Everyone has it?

Okay, then let's start.

Okay, so in the beginning of the last lecture we talked about thresholding, so it was basically

a leftover of the penultimate lecture.

And here the question is what is or are, so also multiple options would be possible, equivalent

discriminant criterion or criteria for thresholding.

So, maybe you remember what the OZL thresholding is maximizing or minimizing and now you can

here choose which ones you think are possible.

Could also be done.

Okay, maybe I can also make it full screen.

Okay, what do you think?

So, what do we want to do?

Do we want to maximize the intra-class variance, minimize the intra-class variance or what

is possible?

Maximizing the inter-class variance and minimize total variance.

So, remember the intra-class variance was the within, so in one class and the inter-class

variance was the one between the classes.

Okay, I think we can 16 people.

Well, I mean today is a bridge day, so not many people are here.

I guess we can see what you say.

Okay, so, maximizing the inter-class variance is certainly right.

This is what you want.

But also, but what you also would like or what would be alternatively correct would

be B, minimizing the intra-class variance would also be fine.

So, B and C would be here correct.

Okay, and minimizing the total variance of course does not work.

So, maximizing the total variance would also work because total variance equals

intra-class variance plus inter-class variance.

So, maximizing would also work but not minimizing.

Okay, B and C here.

So, okay, let's go to the next question.

And I mean this is a quite simple question.

I hope everyone knows what the answer here is.

So, we talked then, now after we talked about thresholding, we went over to

filtering and about convolution then in the end.

And there were some criterion, linearity and shift invariants.

And since they are quite important, I thought let's make a small question here.

Does anyone need still the QR code?

There were two people who were joining the lecture.

No? Everyone has it?

Okay, otherwise it was the code is BQHJ after Stuttgart blah, blah, blah, vote

and then the code.

Okay, 14, we had some more in the last question.