C++/C#/Java video analysis question

Could someone 开发者_如何学Gogive references / frameworks, which allow to do something like Microsoft Kinect does, but using only one video stream?

I would like to see algorithm abstracts and papers if there are some (I hope there are).

If somebody had practical experience of using them, please share it :)

---

Take a look on this project, it implements simple recognition algorithm of the hands gesture using one video stream.

---

Also there is OpenCV. I love this library. A simple google search will yield plenty of info.