ARToolkit – A Neat Computer Vision Tool

Computer Vision allows some really neat things.  ARToolkit is a computer vision library created by Hirokazu Kato used for tracking a surface and placing an animated model on top of it.  It looks really neat.  This summer I’ll be researching more into it and playing around with it.  What you do program a surface into it, and then, using a camera, you can place animated 3D models into your real world environment.  Pretty cool.  If it sounds confusing, don’t worry.  It’s very simple to understand once you see it in action.

Here’s a video showing more details into what it does and how it works. It’s actually a programming library which you can use in Visual C++ and integrate into your applications. From what I can tell, it takes in a video feed from a camera using DirectX, processes the feed images using something like OpenCV (or possibly his own computer vision code, I don’t know), finds the surface and determines its position and orientation, and then renders the 3D model on top of it using OpenGL. I plan to write a lot more posts on computer vision this summer. I’ve been debating between spending my summer doing physics research at MST, or spending all my free time researching quantum mechanics and Computer Vision. Decisions decisions!

And just in case you didn’t notice, the 3D model they’re using is Hatsune Miku, a “vocaloid”. Japanese developers have created technology allowing synthetic voices to sing to music. That’s what you’re hearing in the background. Miku is a sort of virtual “diva” in Japan, and “she” sells lots and lots of records. They even have concerts where Miku is rendered using lasers refracting off of a glass plate.

One of the comments for this video sums up a lot of things,

see while Japan is creating all the cool things, America is busy making bombs and crap to control everyone 🙁

– SailorPerson

Another thing to keep an eye on is new holographic technology coming out.  Japanese scientists just recently made a breakthrough allowing them to make better holographic images using lasers.  You can read about that here.

The team at Osaka took another approach, they use both lasers and white light. They first fire a laser at an object, say an apple, to create an interference pattern, but instead of just one laser color, they actually use three; red, green and blue. The interference pattern is then captured on a light sensitive material which is coated with silver (because it contains electrons that are easily excited by white light) and silicon dioxide (to help steer the waves). They then shine a steady on the metal sheathed material exciting the free electrons, causing the creation of surface plasmons, which results in the regeneration of the captured image as a true-color 3-D ; one that can be viewed from almost any angle and is the same colors as the original object.

Currently, the technique has only been shown to work on still images, and the results displayed on a very small surface area (about as big as a baseball card), but the results of research is nonetheless a very big step towards creating not just more realistic holograms, but true animated 3-D technology.

Here’s an image of an apple which they made using this hologram technology.

The more I read about research Japanese scientists are doing, the more I admire them.  Japan is a beautiful place.  I hope they’re able to recover from their recent earthquake disaster quickly.

Leave a Reply

Your email address will not be published. Required fields are marked *