A OpenCV based augmented reality sandbox implementation. This repository contains the code created by our three person team during this 5 day university project. The basic idea is to sense the height map of a sand surface the user can interact with in the real world and project a corresponding elevation map of the surface right back onto it.
While there are a lot of better implementations of this concept already out there we nevertheless are quite pleased about what we could achieve in 5 days without any previous knowledge of either OpenCV nor image processing. We even had some time left over on the last day to implement a treasure hunting Easter egg in which you can dig for a hidden treasure chest. All in all this was way to much fun ;) Playing with sand all day and getting credits for it.
The code has not been touched after the projects conclusion so expect rough edges.
Unfortunately the images don't really do justice to how vivid the colors look when projected onto the white sand but here are some anyways:
Sandbox with a few mountains and valleys to show off the colorband:
Another picture with the same colorband:
Getting creative with a different colorband:
Yet another example with different colors:
And another one:
Picture of the cropped and undistorted B/W depth field picked up by the sensor.
Picture of the sandbox setup in the lab:
Animation of the auto-calibration method which detects the extent of the projection area as well as distortion. It does so by projecting circles into the edges in a specific order:
- ASUS XTion PRO Live
- In theory any OpenCV/OpenNI compatible sensor like Kinect might do with a bit of work
- Beamer
- Ours had a 1024x768 resolution
- Box with white sand
- We had approximately 9cm of sand
- Some shovels to play with the sand
- OpenCV 2.4.0
- Needs OpenNI support compiled in
- Visual Studio 2010
- Git
In no specific order:
- Aleksej Lednev
- Patrick Hillert <silent@gmx.biz>
- Stefan Hacker <mail@hacst.net>