Hello
I am trying to figure out a way to recognize hyperbola shapes from a given picture
what I want to do is expressed clearly in the picture added :
I dont know if it is really crazy trying to do this approach in grasshopper
thanks
There’s several tons of stuff related with image/pattern recognition around. Just Google it (coding skills : medium to advanced). Most solutions are related to classic Matrix manipulation algos and techniques.
1 Like
Hello,
This is quite a mysterious problem, and without any real image to work on, hard to establish a procedure.
Here is a tool you may find useful, it builds a bunch of hyperbola from two end points and a mid point.
Using math info from this Rational BĂ©zier Curves: Conic Sections
My idea would be to somehow interpolate points among the white zones, and then approximate those curves by hyperbolas…
HyperbolaGrasshopper.gh (26.5 KB)
I don´t know if something like physic simulation could make the work,
I mean maybe… like a fluid simulation, from up to down
at the end is like a kind of cascade that at some points it fall to the sides (when Hyperbolas)
The application is to read radagrams,
antena radar that goes in one direction through soil
That’s rather focused (but the pseudocode of the proposed clustering algorithm appears interesting).
To get a broad idea about what’s happening these days on that matter (the general case) … well … you need to spend several days.
Imagine talking about IPR 30 years ago …
Not all all: as I said that’s a classic Matrix pattern recognition thingy. See some simple/naive tests on that matter (That said IPR is 1++++M miles away from my interests).
The animated thing does a rnd 0/1 density Matrix and then attempts to find/match the required pattern (as Matrix).
Moral: we all live in The Matrix
Looks really amazing Lord Julio…
let me think a bit of it
Some Tequilas from that brand are divine I confess.
Other than that … you’ll need some sort of learning algo: Classic stuff if we have some start pattern (say one of your Curves as a Matrix) and we want to “teach” the idiot (the computer) what else/similar to look for. But things get complicated because we start from some “basic” bitmap resolution and we may refine that if there’s hints around for success. This is (kinda + kinda) the core logic of a cruise missile that emits nothing (it only “sees” a digital interpretation of the given portion of the terrain).
Tip: contact Elon.
1 Like