A Module for Automated Generation of Planar Object Descriptions

Abstract

The article describes the implementation of a module for automated generation of descriptions of planar objects. The module is based on dividing the objects in a scene into parts by using local symmetries, and describing the relations among detected parts. The relations among parts are represented by the knowledge representation scheme based on Petri net theory (KRP). The module is intended to be a part of a computer vision system using the KRP scheme for representing knowledge about objects in the scene.