Catalogue of Artificial Intelligence Techniques
Image Flow Estimation
Keywords: aperture problem, motion constraint equation
Author(s): R.M. Cameron-Jones
The estimation of the two-dimensional velocity field which maps the image intensity pattern at one time instant on to that at the succeeding time instant. The image flow is equivalent to the Optical Flow if the image intensity value for the (changing) image point corresponding to a point in the world remains constant. Methods for recovering image flow are usually based on the motion constraint equation (which relates the local image velocity to local spatial and temporal intensity derivatives) and thus are hindered by the aperture problem--the local irrecoverability of velocity along a direction of zero intensity gradient. For this reason methods for flow recovery at points other than intensity extrema require the imposition of further constraints, and Regularisation methods are commonly used for flow estimation on curves or over regions.
- Nagel, H.H., On the estimation of optical flow: relations between different approaches and some new results
Artificial Intelligence 33 (1987), 299--324.