So far in this chapter, we have gone through some math and we can reconstruct the depth of a scene based on a couple of images taken from different angles, which is a problem of reconstruction of a 3D structure from camera motion.
In computer vision, the process of reconstruction of 3D structures of the scene based on the sequence of images is usually referred to as structure from motion. A similar set of problems is the structure from stereo vision—in reconstruction from stereo vision, there are two cameras, located at a certain distance from each other and in structure from motion, there are different images taken from different angles and positions. There's not much difference conceptually, right?
Let's think about human vision. People are good at estimating distance and relative locations of objects. A person doesn&apos...