Motion- Optical Flow

(parts adapted from http://www.icaen.uiowa.edu)

Optical Flow is an approximation of the 2D motion field. Motion in the world usually occurs in 3D, but, we have a 2D image sensor. So, we see the results as movement accross the 2D image plane. Hence we are seing the projection of the 3D moving points onto the image plane.

Optical Flow Vector - describes motion of each point
vector - image velocity of a point in the image

What can we do with this?

Motion Detection
Motion Compensation
Motion-based Data Compression
3D scene reconstruction
Autonomous Navigation

Can we really do all of this given we have only the projection of the 3D movement?

Well...not really....we need to make some assumptions to accomplish the above, especially when reasoning about 3D (either reconstruction or motion).
For example, one assumption often made is

the changes in image intensities in the image are caused only by motion.

when will this not be true (....illumination sources moving, objects entering/exiting the scene, occlusions)

Optical flow computation

Global and local optical flow estimation

Optical flow computation approaches

Optical flow in motion analysis