A subfield of computer vision where controlled movement of the viewpoint of the imaging camera is an integral part of the image-processing task. Previously in computer vision research, in order to reduce the enormous complexity of visual data, fixed camera geometry and static images have been beneficial in constraining and simplifying the image-processing tasks. Active vision takes a different approach and, by analogy with animal vision, does not avoid movement but gains information from the dynamics of changing viewpoints to resolve ambiguities, gain depth information, and establish relationships between visual sensing and action.