Author(s)

S. Panayiotou & A. Soper

Abstract

Locating a fault or recognising an object in an image requires complex processing normally resulting in a sequence of stages: image enhancement, segmentation of areas of interest and detailed matching of the objects. Each of these stages is normally broken down further into a sequence of standard image processing primitives. Many of these primitives are now commercially available in the form of a library. The total number of such primitives is now large. To successfully locate an object a sequence of these primitives needs to be chosen and their associated parameters tuned. This paper presents a planning system for performing this task. Further the system can adapt to changes in the environment (e.g. lighting, noise and, object) efficiently through r

Keywords