Microsoft Researchers Create 3D Images

Researchers Use Artificial Stereo Vision to Create Mars Images in 3D

"Artificial stereo vision emulates the human ability to see the world in three-dimensions using two eyes. The NASA Web site has provided these shots as an "anaglyph," designed for viewing in 3D with red-green glasses.

"We emulate what the brain does in converting the two views into 3D by a technique called stereo matching. This enables us to produce novel views of the scene, and hence the movie in which the camera seems to move around the 3D object and back again," said Andrew Blake, a senior researcher at Microsoft Research Cambridge.

This technology is being developed to enhance video for teleconferencing in several ways:

- Improved eye-contact
- Environment substitution, which is replacing the environment of the speaker with an artificial one
- Automatic camera pan/tilt/zoom to frame the talker"

Comments

[Bill Kerney]

My wife and I develop distance learning grants for K-12 education (including the BorderLink Project - www.borderlink.org). We are developing an afterschool a high quality fitness training system for kids for sports. We want to use real-time motion analysis to give all kids the high level of support that has formerly been reserved for the world-class athlete.

I am considering having the grant purchase a systems that analyzes human motion but it requires a human operator placing dots on 2 videos of key points of the person in motion. I am looking for a way to get unsupervised markerless input system so we would have no overhead loading up our system. If you have software or ideas that can help please let me know.