I reckon I can work around that using some tech in Quartz I found today.

The red your fingers produce will be much more visible to a cam, especially if filtered, than any other red in the scene.  Somebody would have to point a red light at the surface to fake the blob, which you can do with IR by simply holding a lit match above it…

You could modify touchlib and make a new filter , a red highpass.Whatever is bright red must be a blob , anything else should be discarded.

But a special color filter for touchlib would be usefull in this situation..
Don’t know how you want to preveny applications from showing any strong red ..

I guess no horror movies on the table will be allowed

Maybe you could use some filter in front of the cam which dimms all light a bit, so that the blobs are the brightest light shining through an therefor can be detected by touchlib?

Image attached

Vlado - 30 June 2008 01:23 PM

You could modify touchlib and make a new filter , a red highpass.Whatever is bright red must be a blob , anything else should be discarded.

One way to slove this is the classic way we deal with the noisy signal in the DSP (digital signal processing).

Here is the scheme:

Lets assume that the following steps are done in fast succession and in-sync with the camera capture rate (Firefly MV camera will allow for this).

1. Red Laser off. Capture the camera Image #1. (this is considered to be a noise)
2. Red laser on. Capture the camera Image #2. (this is signal+noise)
Output Image = Image #2 - Image #1 (noise is canceled, thus all is left is a useful signal)
3. Go to step 1

This scheme will give you the output frame rate that is the half of the camera sampling frame rate, but it will elliminate most of the external factors (such as the environmental light and the light from a projector).
By the way, we could use the same technique in the IR based setup to remove the background noise.
This is a real-time (hardware) alternative to a static background filter implemented in TouchLib.

~Alex

whouldn’t this give a problem caused by moving pictures on the screen with changes faster than the half of the cam-rate? espacially if there are red Points or something like this occuring?

Kaiser1401 - 01 July 2008 03:13 AM

Maybe you could use some filter in front of the cam which dimms all light a bit, so that the blobs are the brightest light shining through an therefor can be detected by touchlib?

Would a red filter infront of the camera work do you think? Kinda like the way 3D glasses work...I dunno. This way the intense red light from the laser would be the brightest. Then the mono filter in touchlib should take care of the rest… Just an idea

weyforth - 02 July 2008 09:21 AM

Would a red filter infront of the camera work do you think? Kinda like the way 3D glasses work...I dunno. This way the intense red light from the laser would be the brightest. Then the mono filter in touchlib should take care of the rest… Just an idea

mh, could work i think. ATM i’m just thinking about some options as i don’t (yet) have any hardware to test this. But if it works i’d prefer this setting because 1) the risk of looking into laser light whould be less (as you can see it) and 2) you have the (i think nice) effect of glowing finger tips. 3) You don’t have to modify your webcam.

I honestly think we are over thinking this, when my red lasers arrive I’ll do a test, but I think a simple re-config will work.  Can’t wait, can’t wait, repeat with smile.  Lasers and Lux, I’m so excited.

Kaiser1401 - 02 July 2008 08:46 AM

whouldn’t this give a problem caused by moving pictures on the screen with changes faster than the half of the cam-rate? espacially if there are red Points or something like this occuring?

This is why your camera capture rate must be at least twice as fast as the fastest change on the screen. For example of this, look at the thread about modified version of touchlib and a Firefly MV camera that captures at 112fps.

~Alex