All right, okay, welcome.

I'm very happy that all of you are here.

And today we will discuss the next exercise.

The next exercise is a bit more tricky than the last ones, I would say, because this time

you're working with a paper.

It's an older paper, but you more or less have to re-implement the stuff that was originally

implemented in that one paper.

All right.

And yeah, the paper, it's also uploaded to Stodon.

It's called palm print identification by Fourier transform.

And it's more or less, yeah, to detect the palm prints of different people, to compare

them with one another.

And the basic idea is that you, well, you have very unique fingerprints, but who doesn't

like your hand or your palm prints or whatever aren't unique as well.

So maybe they could be also used as a unique identifier, just as a basic idea.

And to do all of that, there's quite a lot that we have to do or that you have to do

in this case.

Yeah, it's more or less all mentioned here.

So the basic idea is that we have our image or our images, our collection of images,

then we more or less want to extract the center patch to further work with for feature

extraction and so on.

But to make this possible, we have our hand and we have to somehow adapt it, shift it,

move it, rotate it so that we have comparable palm prints.

Because obviously we have these creases.

Obviously you can't see it here from the front, but we have those creases on our hands.

And if we want to compare them, we have to align our hands in some way.

So that's what we're doing before that.

Just as an example, if my hand was like that, we usually want to align it in a way that

we could draw a line across those points here across my fingers.

Could you maybe share your screen?

I'm sorry.

Yeah, I'm sharing my screen.

Completely forgot about it.

You didn't miss anything, luckily yet.

But yeah, sorry about that.

Yeah, that's only the one part that I mentioned.

Again, for the people online, like here, top left, we have our hand.

That's not rotated yet.

We have to rotate it based on, I hope we can see it now, based on this line more or less,

where our fingers more or less meet the palm.

And then we have to rotate our hand.

That's like the first step.

We'll do some more pre-processing, but this is, like I said, the most important step.

And then we want to get the main part of the hand here, as I already mentioned.

Oh, well, that wasn't intentional.

And then we can more or less extract some features by Fourier transforming everything.

And we also will talk about the Fourier transform shortly today.

I will discuss it in a bit more detail next week.

The only thing that you have to know is that you have to use the Fourier transform.

There are 2D models available.