Convert flat images into 3D models

Computer scientists from the university of Stanford have developed an algoritm which they call "Make3d". A suitable name?

Extracting three-dimensional data from still images is an emerging field of computer technology. Make3d does this completely automatically. It takes a two-dimensional image as input, and creates a three-dimensional "fly around" model. This way you can experience the depth of the scene, and view it from numerous locations. If you install VRML viewer or Adobe Shockwave, you can even fly around in your favourite scene.

The algorithm uses a variety of visual cues that humans use for estimating the three-dimensional aspects of a scene. To make a computer perform such a complex task they utilize powerful machine learning techniques. This way the computer can learn the three-dimensional structure of a scene (as a function of the image features).



This is a very interesting step in technology, simply because it's a way to digitalize our world. This digital information opens up doors in many other technologies. The application, the researchers say, could range from enhanced pictures for online real estate sites to quickly creating environments for video games. And also, it might be a great way to improve the vision and dexterity of mobile robots as they navigate through the spatial world.
_______________________________________________________
Further reading, and sources:
Make3d website
Article on science daily