We're trying to build a QC that will take two video sources off of a
server (the easy part) and composite them into one QT movie (another
easy part). The hard part is switching between sources based on
which source is moving. We'll also be taking ONLY the audio from the
speaker camera. So, it really comes down to a way to choose which
video stream to select based on which has more changing pixels.
Subtract successive frames (motion detection), then use a histogram.
If the image is stationary, the low indexes of the histogram will have
high values. If the image is moving, the low indexes will be lower.
You'll have to do some tests to see what thresholds you should use --
I highly recommend testing _In The Environment You'll Be Deploying In_
to get accurate thresholds. smoothing and hysteresis are also good
ideas, if you're feeling ambitious :)
Here's a lame example I whipped up -- no smoothing or anything, but it
might be a start.
Video Input is asynchronous, so instead of subtracting adjacent frames
(t0 - t1), I added a couple frame delay with the queue; this seems to
give better results on my machine since adjacent frames can sometimes
be identical.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartzcomposer-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartzcomposer-dev/email@hidden
This email sent to email@hidden
