Panel summary image interpretation and ambiguities
pp. 319-338
in: Virginio Cantoni (ed), Human and machine vision, Berlin, Springer, 1994Abstract
The goal of the panel is to discuss if ambiguity can be a means for performing cognitive tasks. This position contrasts with the traditional view which considers ambiguous situations as pathological ones to be avoided or recovered as soon as they arise. Evidence exists that the human visual system attempts to process its inputs according to two different representations, the distal and the proximal one. N. Bruno as a cognitive scientist introduces the argument and speculates on the cognitive function served by the proximal interpretation. On the other side, P. Mussio explores - from the point of view of an image interpreter - some situations in which scientists and technicians exploit ambiguous image interpretations to better understand the situation under study. Programs - developed to help them in these interpretation activities-explicitly create and exploit ambiguous descriptions of a same image. However, in many cases ambiguity is still a problem. F. Esposito as an AI scientist explores how to face the intrinsic ambiguity in learning models of visual objects.