It is well known that the different perspectives of our two eyes lead to a slight relative displacement of objects (disparities) in the two monocular views of a scene. The closer an object is to our eyes the larger the relative displacement becomes. We make use of this effect in everyday life to obtain 3-dimensional information about the world surrounding us.

The very same effect can be obtained in machine vision by using two cameras. By placing the two cameras a known distance apart we get a left and right stereo image of the crate containing bottles. The distance we have to shift the images to align them tells us how far away from the camera the objects are in addition to the length and width. These parameters cannot be measured by a single camera viewing the scene from above. Typically the bottles are lower than the crate walls. This means we must make one shift to align the crate and another to align the bottles. From these two estimated disparities we retrieve the crate height and the bottle height.

Images taken with the two cameras in the stereo vision system. There are 5 bottles placed in the right "column" of the crate. The disparities and thereby the height of the objects are estimated using gray scale correlation techniques.