Rule based image segmentation

Abstract

This thesis describes a rule-based approach to segment the image through combining boundary and region analysis. The purpose of this thesis segments the image into background and meaningful region. The duality property of region analysis and boundary analysis is pursued and is utilized for compensating each other``s result. We make a production system to integrate boundary and region analysis. A prototype version is implemented on a VAX 11/750 using KSM image system by the OPS5, FRANZLISP and C program language.