Efficient integration method for human facial images retrieval based on visual content and semantic description

AHMED ABDU ALATTAB; SAMEEM ABDUL KAREEM; IBRAHEEM M.G. ALWAYLE; ANWAR ALI YAHYA; KHALED M.A. ALALAYAH

Author(s)

AHMED ABDU ALATTAB Deppartmen of Computer Science, Faculty of Science and Arts, Sharurah, Najran University, Saudi Arabia; Department of Computer Science, Faculty of Comp. Science and Info. Systems, Thamar University, Yemen.

SAMEEM ABDUL KAREEM Department of Artificial Intelligence, Faculty of Comp. Science and Info. Technology, University of Malaya, Kuala Lumpur, Malaysia.

IBRAHEEM M.G. ALWAYLE Department of Math. Science, Faculty of Science and Arts, Amrran University, Yemen; Department of Computer Science, Faculty of Science and Arts, Sharurah, Najran University, Saudi Arabia.

ANWAR ALI YAHYA Faculty of Computer Science and Information Systems, Najran University, Najran, Saudi Arabia; Department of Computer Science, Faculty of Comp. Science and Info. Systems, Thamar University, Yemen.

KHALED M.A. ALALAYAH Department of Math. & Computer Science, Faculty of Science, IBB University, Yemen; Department of Computer Science, Faculty of Science and Arts, Sharurah, Najran University, Saudi Arabia.

Abstract

A semantic-content based facial image retrieval (SCBFIR) technique that incorporates multiple visual and semantic features to improve the accuracy of the facial image retrieval is proposed. The proposed technique based on reducing the semantic gap between the high-level query requirement and the low-level facial features of the human facial image. Visual features and semantic features are extracted by different methods, moreover, some features may be considered more important than others, so features weighting is used to distinguish the importance of the various features. This research proposed a model that links the high-level query requirement and the low-level features of the human facial image. A newly proposed method based on radial basis function network is introduced for measuring the distance between the query vectors and the database vectors of the different features for finding, weighting, and combining the similarities. The proposed system of SCBFIR is trained and tested on the 'ORL Database of Faces' from AT&T Laboratories, Cambridge, and a local database consisting of local facial images from the University of Malaya (UM), Kuala Lumpur. The results of the experiments show that, as compared to the current content-based facial image retrieval technique (CBFIR), the proposed methods of SCBFIR achieve the best performance. More precisely the CBFIR achieves 84.0% and 92.41% accuracy, while the SCBFIR achieves 97.85 % and 99.39% accuracy for the first and second database respectively within the top 10 retrieved facial images.

Publication Details

Page(s) 1134-1150

DOI DOI not available

Published Journal: Journal of Theoretical and Applied Information Technology, Volume: 98, Issue: 8, Year: 2020

Request the full text from PASTIC

Keywords

Face Retrieval