camera tracking in daylight

May 8, 2007 at 11:03am

camera tracking in daylight

Hello list.

I’m trying to track peoples’ movement with web cam in daylight with jitter.
The trouble I have is that the background is always interfering, instead of
only the foreground (the people).

Has anyone ever tried a heat-seeking camera? I’m assuming that this would
enable a high differentiation between subject and background? Or how about
an infrared camera? It’s important that the sensors (cameras) do not impede
peoples’ movements.

I’m new to this area and would really appreciate your thoughts.

Many thanks.

Jonathan.

#31805
May 8, 2007 at 4:38pm

It seems like you’ve got the right idea. I have never actually worked with interpreting visual data or tracking movement from a camera feed, but it seems like you’d have a few options.

Have you tried tweaking the contrast/brightnss of the video feed to create a better defined edge between ‘performer’ and ‘background’? Can you have your ‘performers’ wear black clothing, so they reflect less light and create contrast more starkly. There are a slew of color-adjustment software tools that might do the trick.

It would also seem that a camera the detects heat wouldbe a sure-fire bet.

what is the setting of your performance more specifically?

#103754
May 8, 2007 at 4:52pm

Some links:

http://www.frieder-weiss.de/eyecon/infrared.html
http://www.hoagieshouse.com/IR/
http://www.rit.edu/~andpph/text-infrared-filter.html

http://world.honda.com/HDTV/IntelligentNightVision/200408/

-Matt

On May 8, 2007, at 12:38 PM, seth wrote:

> It seems like you’ve got the right idea. I have never actually
> worked with interpreting visual data or tracking movement from a
> camera feed, but it seems like you’d have a few options.
>
> Have you tried tweaking the contrast/brightnss of the video feed to
> create a better defined edge between ‘performer’ and ‘background’?
> Can you have your ‘performers’ wear black clothing, so they reflect
> less light and create contrast more starkly. There are a slew of
> color-adjustment software tools that might do the trick.
>
> It would also seem that a camera the detects heat wouldbe a sure-
> fire bet.
>


Matthew Lewis
Advanced Computing Center for the Arts and Design (ACCAD)
The Ohio State University

http://accad.osu.edu/~mlewis

#103755
May 8, 2007 at 6:35pm

there is a object for os x called wiiremote

http://www.iamas.ac.jp/~aka/max/#aka_wiiremote

that handles data from the nintendo wii remote although its only for os x
however the source code its published so a pc version could be made

how??

with the source code its possible to do even a pure data version?

#103756
May 9, 2007 at 12:21am

> Has anyone ever tried a heat-seeking camera? I’m assuming that this would
> enable a high differentiation between subject and background? Or how about
> an infrared camera? It’s important that the sensors (cameras) do not impede
> peoples’ movements.
>

There have been a number of posts on infrared cameras for tracking on the and the Jitter forum.

Also, there is some background subtraction patch in cv.jit that is worth looking at. If only the sun would stop moving.

mz

#103757
May 9, 2007 at 3:52pm

Hi all.

Thank you so much for your helpful replies.

Basically, the people that I want to track are the general public in a park,
so I can’t really ask everyone to wear black :-)

Also, the background might be moving slightly (pigeons or dogs, or
whatever), so obviously therefore the background suppression stuff won’t
work (although I’ve these types of things before in controlled environments
and they worked really well!).

And thanks, Matt for the links which are very helpful. The Honda invention
looks excellent, but I’m guessing it’s either unavailable or very expensive
(or more likely, both!!).

Thanks again.

Jonathan.

On 09/05/07, mzed wrote:
>
>
>
> > Has anyone ever tried a heat-seeking camera? I’m assuming that this
> would
> > enable a high differentiation between subject and background? Or how
> about
> > an infrared camera? It’s important that the sensors (cameras) do not
> impede
> > peoples’ movements.
> >
>
> There have been a number of posts on infrared cameras for tracking on the
> and the Jitter forum.
>
> Also, there is some background subtraction patch in cv.jit that is worth
> looking at. If only the sun would stop moving.
>
> mz
>
> –
> || michael f. zbyszynski — molecular gastronimist
> || http://www.cnmat.berkeley.edu/
> || http://www.mikezed.com/
>
>
>
>


Jonathan Green
0777 1680 497
jonathan@jg1983.co.uk

#103758

You must be logged in to reply to this topic.