What is Stereo Vision?

Humans are able to extract depth information by taking the two views of a scene from each of our eyes. We can even give ourselves two images from two different angles and perceive a 3-D scene. This is the case with binoculars, stereograms, and the famous ViewMaster(tm).

How the brain does this is not the topic of our interest however. We were interested in how to get a computer to figure out the relative depths of objects in a scene given two snapshots from two different angles. To illustrate, consider the following picture of some spheres sitting on the ground, each at a different distance away.

Our eyes can tell which ones are farther away and which ones are closer. But how do we tell a computer to do this? Let's add another view, taken from a position just to the right.

As you can see, the objects in this image are slighty displaced from the picture above. This should give all the information that is necessary for a computer to calculate the depths of each of the objects.

Putting your brain to work

Now to see your brain do this, stare at these two images next to each other for awhile until you can see it in 3D!