this post was submitted on 17 Aug 2024
27 points (93.5% liked)

todayilearned

1173 readers
1 users here now

See !til@lemmy.world

founded 2 years ago
 

not sure why i found this fascinating. i was working a geospatial mapping project and stumbled on this tangent

top 16 comments
sorted by: hot top controversial new old
[–] Beryl@jlai.lu 21 points 4 months ago* (last edited 4 months ago) (1 children)

This paper says nothing about the "brain's navigation system". It is focused on the distribution of cone photoreceptors in the retina, and more precisely in the fovea.

1000017592

They are indeed organized in a hexagonal mosaic as you can see in the picture, and the authors present a new method to estimate spatial distribution of said cones, showing that there are anisotropies, with the cones having a larger local spacing along the horizontal axis.

The fovea (A) is the central-ish region of the retina, and it is packed with cones, which are specialized in color recognition. As you move to the periphery of the retina (B), another type of receptor becomes dominant : the rods. They are a lot better at sensing light but can't tell which color it is. That's why our peripheral vision is mostly shades of gray, even if your brain tries to convince you otherwise by adding the colors from context.

Anyway, not saying this isn't an interesting topic, but it's not the best source to illustrate it.

there was a second reference link for navigation piece:

https://ieeexplore.ieee.org/document/1598543

i took both links from this paper

https://richard.science/sci/2019_barnes_dgg_published.pdf

[–] LazerDickMcCheese@sh.itjust.works 16 points 4 months ago (3 children)
[–] GardenVarietyAnxiety@lemmy.world 4 points 4 months ago

I love that I can count on this being commented whenever hexagons are mentioned 💕

[–] RootBeerGuy@discuss.tchncs.de 1 points 4 months ago (1 children)

The Pentagon would like to know your location.

[–] 5ibelius9insterberg@feddit.org 1 points 4 months ago (1 children)

Go away Pentagon! Why is the Pentagon afraid of the Heptagon? Because the Heptagon octagoned the Nonagon.

[–] Ferris@infosec.pub 1 points 4 months ago (1 children)
[–] 5ibelius9insterberg@feddit.org 3 points 4 months ago* (last edited 4 months ago)

You mean, because he is the

badumm ts

OCTAGONIST ???

[–] SpaceNoodle@lemmy.world 7 points 4 months ago

Find a more efficient packing pattern.

[–] henfredemars@infosec.pub -2 points 4 months ago (1 children)

It makes sense because the real world is isotopic. Rectangular is presumptuous to some extent where high frequency information will be found.

[–] idiomaddict@lemmy.world 2 points 4 months ago (2 children)

I understand each of those words.

[–] henfredemars@infosec.pub 1 points 4 months ago

What’s really cool as that the basis of support in the frequency domain has the same shape as your sampling function.

And generally speaking, the perfect shape would be a circle because you can fit the maximum amount of bandlimited noise into that space. Orientation really shouldn’t matter. It’s stranger that it does.

[–] Randomgal@lemmy.ca 0 points 4 months ago (1 children)

He's just trying to sound smart. Lol

[–] henfredemars@infosec.pub 2 points 4 months ago* (last edited 4 months ago) (1 children)

My dude I have like five publications in the space. But yeah I’m just some guy on the Internet. I just wish I had more opportunities to talk about it.

It started in high school when I had a side job catching dragonflies. They did experiments on their eyes, which have a hexagonal shape. It was being done to research new imaging systems. Later on I worked on some of the mathematical theory for image processing operations, namely how to perform operations like convolution in sampling systems that have non orthogonal basis vectors. Typically you represent these operations using matrix arithmetic but it doesn’t work when your sampling basis is not orthogonal.

I would like to identify those specific operations but I’m pretty much the only guy who would turn up in the search results, so I’m not sure how much more specific I can get. It is unfortunate. To my knowledge I have the fastest convolution implementations for non-rectangular two dimensional sample systems. It’s kind of a lonely research area. It’s a shame.

[–] Randomgal@lemmy.ca 0 points 4 months ago

Cool RP bro.