This volume assesses approaches to the construction of

computer vision systems. It shows that there is a spectrum

of approaches with different degrees of maturity and

robustness. The useful exploitation of computer vision in

industry and elsewhere and the development of the discipline

itself depend on understanding the way these approaches

influence one another.

The chief topic discussed is autonomy.True autonomy may not

be achievable in machines in the near future, and the

workshop concluded that it may be more desirable - and is

certainly more pragmatic - to leave a person in the

processing loop.

The second conclusion of the workshop concerns the manner in

which a system is designedfor an application. It was agreed

that designers should first specify the required

functionality, then identify the knowledge appropriate to

that task, and finally choose the appropriate techniques and

algorithms.

The third conclusion concerns the methodologies employed in

developing vision systems: craft, engineering, and science

are mutually relevant and contribute to one another.

The contributors place heavy emphasis on providing the

reader with concrete examples of operational systems. The

book is based on a workshop held as part of the activities

of an ESPRIT Basic Research Action.