Structural/Syntactic Methods in Perceptual Organization of Shape Modelling and Analysis

The described method for 3D-shape modelling and analysis represents a model-based approach to object representation and recognition, in which model of a 3D-shape is composed of its graph representation which is based on a structural model of a 2D shape. In the presented method laws of perceptual organization (proximity, continuation, symmetry, closure) are applied in the process of primitive detection and grouping of shape features. The constructed feature knowledge base contains models, that means a set of graph representations of objects together with the minimal representations of their sweep contours in the form of context-free grammars designating them. During perceptual modelling of a 3D object its graph representation is constructed automatically which enables automated analysis and recognition of a modelled 3D object.