AN ATTRIBUTED GRAPH STRUCTURE FOR OBJECT RECOGNITION

摘要

Since graphs are very versatile data structures, they can be applied for various computer vision problems. The first part of the paper presents an overview of the field of graph theory in object recognition. The second part presents a new graph structure, which is better suited for representing parameterized image features. A feature class is assigned to every node, whose parameter vectors are stored in matrix format. Finally a simulation is presented to demonstrate the applicability of the method.