Search and retrieval of three-dimensional media will rapidly become a key issue in the upcoming panorama of multimedia content: 3D models are indeed expected to represent a huge amount of traffic and data stored in the Internet. This article proposes a novel technique to define and construct 3D shape prototypes, that improve the automatic classification of 3D content. The shape-prototype summarizes the most relevant features of the members of a class. The query object is then classified into the class represented by the prototype more similar to the given query. In the proposed methodology, each member of a class is represented by a structural descriptor encoded as an attributed graph. The prototype is obtained by applying graph-transformation techniques among the shape descriptors associated to the members of the class. The effectiveness of the classification process is finally evaluated on an heterogeneous benchmark of 3D objects. © 2007 IEEE.
Structural Shape Prototypes for the Automatic Classification of 3D Objects
Marini S;Spagnuolo M;Falcidieno B
2007
Abstract
Search and retrieval of three-dimensional media will rapidly become a key issue in the upcoming panorama of multimedia content: 3D models are indeed expected to represent a huge amount of traffic and data stored in the Internet. This article proposes a novel technique to define and construct 3D shape prototypes, that improve the automatic classification of 3D content. The shape-prototype summarizes the most relevant features of the members of a class. The query object is then classified into the class represented by the prototype more similar to the given query. In the proposed methodology, each member of a class is represented by a structural descriptor encoded as an attributed graph. The prototype is obtained by applying graph-transformation techniques among the shape descriptors associated to the members of the class. The effectiveness of the classification process is finally evaluated on an heterogeneous benchmark of 3D objects. © 2007 IEEE.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.