Computer Vision: Science and Technology

The scientific discipline of computer vision encompasses the theory and technology of machine vision as well as computational models of biological vision systems. The European Computer Vision Network constitutes an interdisciplinary group of research organisations and industries with overlapping coverage of each of the above areas.

Computer vision systems have at least three major types of functions :

a)

Control of activity, in particular, vision-guided movement, including movement of sensors or effectors. Examples are locomotion, manipulation, and tracking. This area includes the use of vision for automation.

b)

Interpretation of static, dynamic and multi-dimensional images :
Such interpretation tasks include :

(i)

specialized recognition or labelling tasks such as satellite image interpretation, quality control and medical image analysis,

(ii)

scene understanding for security, surveillance and traffic monitoring.

Man-machine communications : Vision may potentially provide a rich new communication mode through such techniques as understanding of gestures and facial expressions (gaze direction, lip movements etc.), as well as the recognition of users.

The widespread application of computer vision has been held back by overwhelming requirements in computing power and memory. Within the last few years, micro-processors with sufficient computing power and memory have become commonly available at steadily decreasing prices. With this advance in computer power, researchers have shown that it is now possible to construct continuously operating real-time vision systems. Widespread application of such systems requires the development of special purpose hardware, and the development of methods and standards in systems design and systems integration. A base of supplier industries must take up these techniques and provide mass production of reliable components. A cadre of trained engineers must be able to apply these techniques to the needs of end users.