: [1]
: IR blob tracking  ( 6560 )
0 and 1 Guest are viewing this topic.
Karma: +0/-0
: 1
Offline Offline
« : January 27, 2008, 05:48:20 PM »

Hey all,

I have a two-part question about the way in which the Wiimote tracks IR points. In Johnny's finger tracking video, he first waves his hand in front of the Wiimote and IR light array to demonstrate what he describes as "extra noise". What I would have expected to see is a grayscale image of his hand but what we ACTUALLY see is group of blobs. It appears that the Wiimote (or whatever program Johnny ran with it) was quantizing the image into several unique blobs! (in real time, no less) So, first question: Does the Wiimote do this? (uber contrast settings or somthing), or does the accompanying software do it? (any info on what method it uses would be awesome too)

Second question: I keep reading that each Wiimote can track up to 4 points but in the same video of Johnny's hand, we see at least twice as many points and it seems like they're ALL being tracked! What's the deal with that? Doth mine eyes deceive me?

Ok, I have a third question: For the head tracking/3D desktop project, how does the program differentiate between the user twisting his or her head from side to side from the user moving away from the screen. From the viewpoint of the Wiimote, the apparent varying distance between two IR blobs that occurs when the user steps back could be misinterpreted by turning one's head away from the screen. Would adding a third point not solve this problem? How do-able is that?

Any light (visible, not IR. yuk yuk... groan) that anyone can shed on this for me would be greatly appreciated. This is such a wonderfully geeky project. You guys rock.

Karma: +0/-0
: 8
Offline Offline
« #1 : January 28, 2008, 01:14:02 PM »


1. The WiiMote has a 4 point tracker implemented in hardware. This means any image analysis is done by hardware and thus in realtime. I don't know what algorithm is behind this so I can only say the remote picks from alle IR light sources 4 points which it displays. My assumption is that it picks the brightest of all points. So there you have your "blobs".
The WiiMote gives transmits the position and a approximate radius of each point it tracks. That's why the points in the video differ in their size.

2. The WiiMote, according to Johnny, track points with 100Hz. I think in the Video it appears to track more than 4 points because the video only shows 25Hz an thus four full tracking cycles. So not your eyes but the video decieve you =)

3. It does not. The program simply calculates the distance between the points and based on this, the distance of the user's head. And yes it yould be misinterpreted but so can all pointer calculation the Wii does in any game as the sensor bar is also only 2 IR points. Adding a third point could solve the problem but will raise others. See http://www.wiimoteproject.com/wiimote-desktopvr/improving-to-6-dof-head-tracking-t65.0.html for more information.

Google is your friend.


: [1]

Clicky Web Analytics