| K. Raftopoulos, P. Tzouveli, K. Ntalianis, S. Kollias, D.Kalomoirakis, N. Fyssas, G. Foukaneli |
| Saint Classification in Byzantine Art |
| 14th International Conference on Virtual Systems and MultiMedia VSMM'08, Cyprus, 20-26th October, 2008 |
|
ABSTRACT
|
| Over the last decade several initiatives were carried out worldwide, towards preservation and promotion of cultural heritage. In this framework, many Byzantine artworks of historic, cultural and artistic interest have been digitized, producing large volumes of visual data, and thus motivating the need for effective indexing and retrieval from such repositories. This paper proposes a decision making framework for content-based retrieval of Byzantine icons, based on a combination of low-level features. In particular initially characteristic icons are selected and a features model is built for the entity under consideration. Next head areas are unsupervisedly detected using Hough transform and the special metrics system of Byzantine art, described by Dionysios from Fourna. Next for each head area proper features, such as MPEG-7 descriptors for shape and color, are extracted. A comparison is then performed between four standard classification algorithms before the decision is made in favour of the multilayer-perceptron with back propagation learning (NNBP). The analysis and results of this classifier indicate the effectiveness, generality, and flexibility of the proposed scheme
|
| 09 September, 2008 |
| K. Raftopoulos, P. Tzouveli, K. Ntalianis, S. Kollias, D.Kalomoirakis, N. Fyssas, G. Foukaneli, "Saint Classification in Byzantine Art", 14th International Conference on Virtual Systems and MultiMedia VSMM'08, Cyprus, 20-26th October, 2008 |
[
BibTex] [
Print] [
 Back] |
