Custom ontologies for an automated image annotation system

Authors:
Gabriel Mihai;Liana Stanescu;Dumitru Dan Burdescu;Marius Brezovan
Affiliations:
University of Craiova, Faculty of Automation, Computers and Electronics, Craiova, Romania;University of Craiova, Faculty of Automation, Computers and Electronics, Craiova, Romania;University of Craiova, Faculty of Automation, Computers and Electronics, Craiova, Romania;University of Craiova, Faculty of Automation, Computers and Electronics, Craiova, Romania
Venue:
KES'11 Proceedings of the 15th international conference on Knowledge-based and intelligent information and engineering systems - Volume Part I
Year:
2011

Citing 6
Cited 0

A translation approach to portable ontology specifications

Knowledge Acquisition - Special issue: Current issues in knowledge modeling
Object Recognition as Machine Translation: Learning a Lexicon for a Fixed Image Vocabulary

ECCV '02 Proceedings of the 7th European Conference on Computer Vision-Part IV
Automatic image annotation and retrieval using cross-media relevance models

Proceedings of the 26th annual international ACM SIGIR conference on Research and development in informaion retrieval
Modeling annotated data

Proceedings of the 26th annual international ACM SIGIR conference on Research and development in informaion retrieval
MPEG-7 as a Metadata Standard for Indexing of Surgery Videos in Medical E-Learning

ICCS '08 Proceedings of the 8th international conference on Computational Science, Part III
The segmented and annotated IAPR TC-12 benchmark

Computer Vision and Image Understanding

Quantified Score

Hi-index	0.00

Visualization

Abstract

Automated annotation of digital images is a challenging task being used for indexing, retrieving, and understanding of large collections of image data. Several machine learning approaches have been proposed to model the existing associations between words and images. Each approach is trying to assign to a test image some meaningful words taking into account a set of feature vectors extracted from that image. In general for the annotation process of medical or natural images the words are retrieved from a controlled vocabulary or from an ontology. This paper presents an original approach for creating two ontologies and an original design of an image annotation system. The ontologies are created using the information provided by two distinct sources: MeSH - a vocabulary used for subject indexing and searching of journal articles in the life sciences and SAIAPR TC-12 Dataset - a set of annotated images having a vocabulary with a hierarchical structure. The annotation system is using an efficient annotation model called Cross Media Relevance Model each image being segmented using a segmentation algorithm based on a hexagonal structure.