INEXACT REASONING IN EXPERT SYSTEMS FOR IMAGE ANALYSIS

It is an inevitable task for expert system to deal with uncertainty in image ana lysis.A variety of alternative frameworks now exists for representing and reasoningabout uncertainty.Among the most prominent are Bayesian probability theory,belieffunctions,and possibility theory.Experts in this field are very solicitous of non-numerical methods of inexact reasoning,such as non-monotonic logic.This paper givesa tentative survey of approaches in applying these inference theories to some particular aspects of image analysis.