This paper introduces the Bag of Graphs (BoG), a Bag-of-Words model that encodes in the graphics of the local structure of a digital object. We present formal definitions, introduce concepts and rules that make this model flexible and easy to customize for various applications. We define two BoG-based methods of Bag of Singleton Graphs (BoSG) and Bag of Visual Graphs (BoVG), which create vector representations for graphics and images. We evaluated the Bag of Singleton Graphs (BoSG) for graph classification on four datasets of the IAM repository, obtaining significant results in accuracy and execution time. The Bag of Visual Graphs (BoVG) method is evaluated for image classification on Caltech and ALOI datasets, and remote sensing image classification on Monte Santo and Campinas dataset images. This framework opens the possibilities for classifying, retrieving, and grouping tasks on large datasets that use impractical graphical representation before due to inappropriate matching graphic complexity.
No comments:
Post a Comment