I now have a stereo vision demo using both quickcams under redhat linux 8.0. I've played around with various stereo algorithms before, but the most successful to date seems to be the one developed by Stan Birchfield. I've modified Stan's original algorithm to include segmentation information which gives a cleaner result.
The performance of the quickcams under linux is very slow (approx 1 frame per second) but this particular algorithm doesn't need to make use of motion information. Although the algorithm is quite computationally intensive by far the majority of the time is taken simply waiting for the images to arrive from the cameras. Check it out at:
I now have a stereo vision demo using both quickcams under redhat linux 8.0. I've played around with various stereo algorithms before, but the most successful to date seems to be the one developed by Stan Birchfield. I've modified Stan's original algorithm to include segmentation information which gives a cleaner result.
The performance of the quickcams under linux is very slow (approx 1 frame per second) but this particular algorithm doesn't need to make use of motion information. Although the algorithm is quite computationally intensive by far the majority of the time is taken simply waiting for the images to arrive from the cameras. Check it out at:
http://www.fuzzgun.btinternet.co.uk/rodney/vision.htm
- Bob