previous up next
Previous: Defintion of Active Vision Up: Origins of Active Vision Next: Computer Vision Systems

Origins of Active Computer Vision

During the last few years, there has been a growing interest in the use of active control of image formation to simplify and accelerate scene understanding. Basic ideas which were suggested by [Bajcsy 88] and [Aloimonos et al. 87] has been extended by several groups. Examples include [Ballard 91], and [Eklundh 92]. Brown [Brown 90] has demonstrated how multiple simple behaviours may be used for control of saccadic, vergence, vestibulo-ocular reflex and neck motion.

This trend has grown from several observations. For example, Aloimonos and others observed that vision cannot be performed in isolation. Vision should serve a purpose [Aloimonos 87], and in particular should permit an agent to perceive its environment. This leads to a view of a vision system which operates continuously and which must furnish results within a fixed delay. Rather than obtain a maximum of information from any one image, the camera is an active sensor giving signals which provide only limited information about the scene. Bajcsy [Bajcsy 88] observed that many traditionally vision problems, such as stereo matching, could be solved with low complexity algorithms by using controlled sensor motion. Examples of such processes were presented by Krotkov [Krotkov 90]. Ballard [Ballard 88] and Brown [Brown 90] demonstrated this principle for the case of stereo matching by restricting matching to a short range of disparities close to zero, and then varying the camera vergence angles. The development of robotic camera heads has lead to the possibility of exploiting controlled sensor motion and control of processing to construct continuously operating real time vision systems. This paper is concerned with the design of such a system.

References:

[Aloimonos 87] J.Y. Aloimonos, I. Weiss and A. Bandopadhay, "Active Vision", International Journal on Computer Vision, pp. 333-356, 1987.

[Bajcsy 88] R. Bajcsy, "Active Perception", IEEE Proceedings, Vol 76, No 8, pp. 996-1006, August 1988.

[Ballard 88] Ballard, D.H. and Ozcandarli, A., ÒEye Fixation and Early Vision: Kinematic DepthÓ, IEEE 2nd Intl. Conf. on Comp. Vision, Tarpon Springs, Fla., pp. 524-531, Dec. 1988.

[Ballard 91] D. Ballard, "Animate Vision", Artificial Intelligence, Vol 48, No. 1, pp. 1-27, February 1991.

[Brown 90] C. Brown, "Prediction and Cooperation in Gaze Control", Biological Cybernetics 63, 1990.

[Eklundh 92] J.O. Eklundh and K.Pahlavan, Head, "Eye and Head-Eye System", SPIE Applications of AI X: Machine Vision and Robotics, Orlando, Fla. April 92.

[Krotkov 87] Krotkov, E., "Focusing", International Journal of Computer Vision, 1, p223-237(1987).

[Krotkov 90] Krotkow, E., Henriksen, K. and Kories, R., "Stereo Ranging from Verging Cameras", IEEE Trans on PAMI, Vol 12, No. 12, pp. 1200-1205, December 1990.


Patrick Reignier
Fri Jun 30 18:47:32 MET DST 1995