PDA

View Full Version : Whats the openGL different from openCV?



cnreader
05-11-2007, 06:43 PM
Can anyone out there answer me the question:
Whats the openGL different from openCV?
which one is more population? and easy to use in practice?

Korval
05-11-2007, 07:01 PM
Whats the openGL different from openCV?Um, they do entirely different things.

OpenGL is a 3D graphics rendering library. OpenCV is a computer vision library "focused on real-time image processing." GL creates images, CV processes them.

cnreader
05-11-2007, 07:07 PM
Thank you very much for your prompt reply and wonderful reply.
I know their difference now.

cnreader
05-11-2007, 07:11 PM
further more, Could I comprehensive as this?
we use the GL creates the image and then we use CV to modify or process the image if we find it not very good enough or satisfact with your requirement?

Korval
05-11-2007, 07:23 PM
further more, Could I comprehensive as this?
we use the GL creates the image and then we use CV to modify or process the image if we find it not very good enough or satisfact with your requirement?You misunderstood.

OpenCV is for doing Computer Vision (http://en.wikipedia.org/wiki/Computer_vision) work.

If you're trying to produce 3D images, you use OpenGL. If you have real-time data coming through a camera and want to make the computer understand the imagery, you use OpenCV.

If you already have a 3D scene in the computer, and you want to make a computer understand it, you use neither. You simply have the computer sample the world in its native internal data structures. So it makes no sense to render to OpenGL and then use OpenCV to process it.

cnreader
05-12-2007, 01:25 AM
sorry I may be not expressing exactly.
you are right, I understand.
What I want to say above is
we can use the openGL to construct a 3D object, for example, a box, then we save it as a jpg file.
then open it with our computer, if we want to modify it into a cylinder by computer itself, we can use openCV to recognise it, for example, use edge recognize method to get its coordenates parameters and change these parameters to form a cylinder.
does this describe right?
-----
btw, due to some politic cause, we cannt access that wikiedia web site.

ZbuffeR
05-12-2007, 01:58 AM
Originally posted by cnreader:
btw, due to some politic cause, we cannt access that wikiedia web site. Too bad for you and your country... Wikipedia is one of the best thing that happened on the web after 2000.
Maybe you can try to read wikipedia, by starting from this page :
https://secure.wikimedia.org/wikipedia/en/wiki/Main_Page

Back on topic :
yes of course you can train your computer vision algorithm on synthetic images generated with OpenGL.


See for ex. this project :
http://ai.stanford.edu/~asaxena/rccar/
Specifically page 4 of the pdf here :
http://ai.stanford.edu/~asaxena/rccar/ICML_ObstacleAvoidance.pdf

cnreader
05-12-2007, 03:00 AM
oh, several years ago, I can access to the web site, and I have ever publish some items in it and later, we found we cannt access to it. some peopel said it was block by our isp, I dont know if its right, but we cannt access, occassionally we could access. forget it!
I know the concept of machine vision. but i need learn too much aswell.

Korval
05-12-2007, 12:27 PM
we can use the openGL to construct a 3D object, for example, a box, then we save it as a jpg file.
then open it with our computer, if we want to modify it into a cylinder by computer itself, we can use openCV to recognise it, for example, use edge recognize method to get its coordenates parameters and change these parameters to form a cylinder.You should never do anything like this.

If you want to create a box and save it, you save the box, not a JPEG image of the box. There are innumerable formats for saving vertex data, from 3DS to COLLADA.

That way, when it comes time to modify the data, you simply modify the data.

V-man
05-13-2007, 12:11 AM
I'm guessing that's suppose to be the point of the exercices : save as jpeg, then recognize it.

If OP needs help with OpenCV, join the opencv mailing list and forums.

cnreader
05-14-2007, 07:02 PM
ok, thank you all, I think Ihave learned very much from your all threads. although I hve some confusion. I can comprehend little by little from basical concept.
btw, V-man, where is the copencv mailing list and forum?

ZbuffeR
05-15-2007, 04:56 AM
http://tech.groups.yahoo.com/group/OpenCV/
Mail to OpenCV-subscribe@yahoogroups.com or use your Yahoo acount.

cnreader
05-16-2007, 12:27 AM
I know this group, I was member of the group, what I want to know is if there is some web site like this one?
too many subscriber out there