Home // ICCGI 2018, The Thirteenth International Multi-Conference on Computing in the Global Information Technology // View article

Using Online Manifold Learning for Color Image Quality Assessment

Authors:
Meiling He
Mei Yu
Hua Shao
Hao Jiang
Gangyi Jiang

Keywords: Color image quality assessment; visual saliency; manifold learning

Abstract:
The structure of the low-dimensional characteristics of images is manifold, which is precisely what the human visual system perceives. With this inspiration, a new image quality assessment (IQA) metric called online manifold learning based quality (OMLQ) is proposed for color IQA in this paper. Online manifold learning is employed to construct a feature extraction matrix, which is used to obtain low-dimensional manifold vectors. In addition, visually important regions are detected to mimic the properties of the visual perception. The new IQA score is defined as the similarity of feature vectors between reference image and corresponding distorted one. Extensive experiments performed on three publicly available benchmark databases demonstrate that the proposed IQA index OMLQ works better in terms of prediction accuracy than some state-of-the-art indices.

Pages: 21 to 24

Copyright: Copyright (c) IARIA, 2018

Publication date: June 24, 2018

Published in: conference

ISSN: 2308-4529

ISBN: 978-1-61208-641-5

Location: Venice, Italy

Dates: from June 24, 2018 to June 28, 2018