View Single Post
daperl's Avatar
Posts: 2,427 | Thanked: 2,986 times | Joined on Dec 2007
#2872
Originally Posted by Capt'n Corrupt View Post
Yes, this would work. I would include interpolation (perhaps this is what you meant by 'smooth the movement') between frames and set a best-guess based on these accumulated values at some point beyond the rendered frame -- though close enough not to produce discernible lag or strange 'drifting'. I would definitely use the acceleration delta of many of these to help determine the best-guess. Linear interpolation would be acceptable due to the minuscule time-interval between guesses. I would also determine some sort of weighted average based on the differing values to best determine where the device is related to the head and tick at regular intervals.
Exactly right, you could use a well known curve fitting (and predicting) algorithm like least squares. Very fast on these devices when only using about 25 data points per second.

I wonder what the power consumption would be for something like this. The camera would have to be running, but I suppose if the majority of tracking code were DSP, it could be done quite efficiently. My only concern would be the facial recognition.

I expect darker faces would also be more challenging due to the reduced contrast.
These look like two good links from the Viola-Jones object detection framework page.

http://www.cmucam.org/wiki/viola-jones
http://opencloovision.codeplex.com/

Not better than sex, but there's nothin' like a good algorithm cook book!
__________________
N9: Go white or go home
 

The Following 3 Users Say Thank You to daperl For This Useful Post: