View Full Version : OpenML - what is it exactly?

04-12-2004, 06:04 PM
I've looked at the info on the OpenML website but I still have a question and feel free to flame me if I've misunderstood what OpenML is.

What I'm looking for is interface capability to a video stream. Say, I connect a TV input into my video card, can I use OpenML to display the video and be able to click in the video window and have OpenML return to me the click coordinates? That is, if I wanted to track an object on the screen (in the video). Oh and can you draw on top of this video window (say draw an overlay)?

I'd appreciate any info..



04-16-2004, 08:51 AM
You can use the ML API to stream video from your video card (***uming it is OpenML-compliant) to your application. You can then display the video to a window, as you would display any image.

Clicking in the window, and drawing overlays, would be handled by the normal UI mechanisms in your OS -- eg, Win32/GDI on Windows. This is independent of OpenML.

In order to track an object in the video, you would need to implement image analysis algorithms in your app.

04-16-2004, 02:13 PM
What does "Open ML compliant" mean ???

I currently detect video devices in a wxWidgets-application with a vfw-wrapper, and that works fine for webcams (this is what have tried so far) and I use Portaudio for getting audio Input.
Will bring Open ML as replacement of the solutions above any benefits to me ? Does it support DV ?

When will it be released ? (The Betaversion did only work the first time I tried it, afterwors not anymore, I don't know why)


04-19-2004, 09:24 AM
<BLOCKQUOTE><font size="1" face="Arial, Verdana, Helvetica, sans-serif">quote:</font><HR>Originally posted by rsaccon:
What does "Open ML compliant" mean ???<HR></BLOCKQUOTE>

It means that there are OpenML hardware drivers for the device.

04-19-2004, 12:18 PM
Given that OpenML is "based/related to" on OpenGL (unless I'm wrong about this)... can I use it together with OpenGL? That is can I have it run in an OpenGL (or GLUT) window and hence have my click callbacks available to me?

Basically, this is what I'm doing: I work for NASA Johnson Space Center in the Visiting Vehicles Office. We support (in particular) the Russian vehicles docking to the space station. Unfortunately the Russians don't give us much data (like the xyz location of the approaching vehicle). The provide us with range and range rate. Using cameras on the space station and the range data we can figure out (calculate) where the vehicle is in xyz coordinates by clicking in a graphic window that represents the video screen. Of course this is quite hard to get down accurately because the video is on one monitor and the graphic window is on our computer monitor...so we need a way to send the video into the computer and also put it into a window that we can click in directly over the image to get the actual vehicle position in the 2D pane (which is then coverted by simply math rotations into the location in 3D space).

Having overlays display on the screen would be nice too....but not a necessary.

If figured if OpenML and OpenGL are closely tied then I can simply use the GLUT mouse functions glutMouseFunc(), etc

can i?



04-21-2004, 09:24 AM
You could transfer the video stream from a OpenML capture card to memory, using ML. You could then take the images and use them as texture maps for an OpenGL surface -- thus integrating the video into your OpenGL rendering environment.

04-27-2004, 05:41 AM
yes, but can you do that in real time...that sounds kind of slow to me.

I figure if you can have a video stream displayed that you should be able to click on it and return the mouse location of the click.
I've been able to achieve a similar thing in Windows using Java with JMF (Java Media Framework) and my Geforce3 V8200 Deluxe capture card...now I'm trying to replicate it under Linux.