Abstract
The paper describes a graph theoretic method for recognizing the color images of natural scenes. Images are represented by an adjacency graph description and the scenes are modeled by means of 3D to 2D constraints and adjacency relations. The relational distance measure is used for matching the relational graphs of the scenes and the respective images. Experiments are conducted on 193/spl times/232 imperfect color images of outdoor scenes involving complex shaped objects and irregular textures. Matching results are presented for one image processed by the method.