- 27 March:
BALTAN Tracker Research Update
Concept: Geert Mul
Programmer: Erik Overmeire
Thermographic motion-tracking system
In my work I started to use a Thermographic camera for motion-tracking instead of a normal camera because the thermographic camera ‘sees’ only heat and no light. This filters out a lot of computer vision misinterpretations that occur when working with a normal camera, where shadows or objects can be mistaken for persons.
September 2009:
Erik Overmeire and myself installed a test set-up at BALTAN Laboratories. I’m using a FLIR VSR-6 camera. One of the reasons is that this camera has a nice wide-angle lens that is often useful. Another reason is that the camera can be used outdoors. We are using Max-MSP both to control the camera, and for motion-tracking. As for the computer, we bought a Pear-C, a mac-clone, basically a PC with OSX running on it. I have to feed the analogue signal into the computer and I’m using an ADVC-55 for that. The results stemming from the development of the BaltanTracker will not only be the tracker itself but also a library of abstractions that could assist in making motion tracking a lot easier for artists working with Max/MSP.
Erik Overmeire has been working on the tracking software. We are talking to the camera in its own protocol using Max/MSP and a serial cable to set the automatic gain, flip the image, etc.
Erik used the CV.JIT (http://www.iamas.ac.jp/~jovan02/cv/) tools to make the tracking patches. Below is an example of the camera output (camera on the ceiling looking down to persons working on a laptop) and how its being processed by the cv.jit patches. On the upper left corner is the original camera output. On the right of that, the same with markers for tracked objects. On the below left corner, the subtracted image, on the right of that, the individual tracked objects marked in different colours.
Erik has also made help files. Below, for example, is a help file for a background subtraction patch.
To Do:
Enabling the OSC communication. This will be also the way to talk to the BaltanPlayer (realtime programmable modular HD multichannel video-player) that is being developed by Telcosystems here at BALTAN. We also have to work out the whole protocol for communication between the BaltanTracker and the BaltanPlayer. So once Telcosystems have developed the BaltanPlayer, we can go ahead with that.
To Make:
Two new concepts related to tracker technology developed during the research. One is an audio/visual sequencer for use in (semi) public space:
The other is a concept for a ‘see-through walls’ set-up, in public space. The idea is to use the tracker to project a spotlight on a wall, following the people that are passing in front of that wall. What will be visible in the spotlight is actually what is behind that wall. This could work either in realtime, or with pre-recorded images. It will make use of the Max-MSP patches that have been developed at BALTAN Laboratories by Mr. Stock Interfaces for Michelle Teran’s installation The City is Creative presented during the flux-s festival in September 2009.

