A vision architecture for integrated manufacturing systems

This thesis introduces a conceptual multi-layered reference model to deal with representing information in a visual scene and to relate this data to similar data structures used within manufacturing processes. In this model, information is processed by a set of algorithms organised within layers of hierarchy. The bottom layer deals directly with data collected by the sensors and the top-most layer offers an interface to the different related manufacturing processes. The building elements of the model and the functionality of each layer are examined with respect to different potential implementation. Through a study of typical vision problems, the thesis shows how conventional image processing solution can be re-implemented in a compatible way with the reference model with the advantage of making use of standard algorithms and development tools. The thesis also examines the impact of using machine vision, based on layered model with other manufacturing entities, and the resulting consequences on computer-integrated manufacturing strategy.