The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
L'algorithme de classement multiple a été adopté avec succès dans la recherche d'images basée sur le contenu (CBIR) ces dernières années. Cependant, alors que les fonctionnalités globales de bas niveau sont largement utilisées dans les systèmes actuels, les fonctionnalités basées sur les régions ont reçu peu d'attention. Dans cet article, un nouveau cadre transductif axé sur l'attention et basé sur une représentation graphique hiérarchique est proposé pour la récupération d'images basée sur les régions (RBIR). Cette approche peut être caractérisée par deux propriétés clés : (1) Puisque la question de l'importance des régions est le problème clé de la recherche basée sur les régions, un modèle d'attention visuelle est choisi ici pour mesurer l'importance des régions. (2) Une représentation graphique hiérarchique qui combine les similitudes au niveau de la région avec celles au niveau de l'image est utilisée pour la méthode de classement multiple. Une nouvelle fonction d'énergie de propagation est définie, qui prend en considération à la fois les caractéristiques visuelles de bas niveau et l'importance régionale. Les résultats expérimentaux démontrent que l'approche proposée montre des performances de récupération satisfaisantes par rapport aux méthodes de classement multiple basées sur le global et sur les blocs.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Song-He FENG, De XU, Bing LI, "Combining Attention Model with Hierarchical Graph Representation for Region-Based Image Retrieval" in IEICE TRANSACTIONS on Information,
vol. E91-D, no. 8, pp. 2203-2206, August 2008, doi: 10.1093/ietisy/e91-d.8.2203.
Abstract: The manifold-ranking algorithm has been successfully adopted in content-based image retrieval (CBIR) in recent years. However, while the global low-level features are widely utilized in current systems, region-based features have received little attention. In this paper, a novel attention-driven transductive framework based on a hierarchical graph representation is proposed for region-based image retrieval (RBIR). This approach can be characterized by two key properties: (1) Since the issue about region significance is the key problem in region-based retrieval, a visual attention model is chosen here to measure the regions' significance. (2) A hierarchical graph representation which combines region-level with image-level similarities is utilized for the manifold-ranking method. A novel propagation energy function is defined which takes both low-level visual features and regional significance into consideration. Experimental results demonstrate that the proposed approach shows the satisfactory retrieval performance compared to the global-based and the block-based manifold-ranking methods.
URL: https://global.ieice.org/en_transactions/information/10.1093/ietisy/e91-d.8.2203/_p
Copier
@ARTICLE{e91-d_8_2203,
author={Song-He FENG, De XU, Bing LI, },
journal={IEICE TRANSACTIONS on Information},
title={Combining Attention Model with Hierarchical Graph Representation for Region-Based Image Retrieval},
year={2008},
volume={E91-D},
number={8},
pages={2203-2206},
abstract={The manifold-ranking algorithm has been successfully adopted in content-based image retrieval (CBIR) in recent years. However, while the global low-level features are widely utilized in current systems, region-based features have received little attention. In this paper, a novel attention-driven transductive framework based on a hierarchical graph representation is proposed for region-based image retrieval (RBIR). This approach can be characterized by two key properties: (1) Since the issue about region significance is the key problem in region-based retrieval, a visual attention model is chosen here to measure the regions' significance. (2) A hierarchical graph representation which combines region-level with image-level similarities is utilized for the manifold-ranking method. A novel propagation energy function is defined which takes both low-level visual features and regional significance into consideration. Experimental results demonstrate that the proposed approach shows the satisfactory retrieval performance compared to the global-based and the block-based manifold-ranking methods.},
keywords={},
doi={10.1093/ietisy/e91-d.8.2203},
ISSN={1745-1361},
month={August},}
Copier
TY - JOUR
TI - Combining Attention Model with Hierarchical Graph Representation for Region-Based Image Retrieval
T2 - IEICE TRANSACTIONS on Information
SP - 2203
EP - 2206
AU - Song-He FENG
AU - De XU
AU - Bing LI
PY - 2008
DO - 10.1093/ietisy/e91-d.8.2203
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E91-D
IS - 8
JA - IEICE TRANSACTIONS on Information
Y1 - August 2008
AB - The manifold-ranking algorithm has been successfully adopted in content-based image retrieval (CBIR) in recent years. However, while the global low-level features are widely utilized in current systems, region-based features have received little attention. In this paper, a novel attention-driven transductive framework based on a hierarchical graph representation is proposed for region-based image retrieval (RBIR). This approach can be characterized by two key properties: (1) Since the issue about region significance is the key problem in region-based retrieval, a visual attention model is chosen here to measure the regions' significance. (2) A hierarchical graph representation which combines region-level with image-level similarities is utilized for the manifold-ranking method. A novel propagation energy function is defined which takes both low-level visual features and regional significance into consideration. Experimental results demonstrate that the proposed approach shows the satisfactory retrieval performance compared to the global-based and the block-based manifold-ranking methods.
ER -