Open AccessQuaternion wavelet transform based full reference image quality assessment for multiply distorted images
Author(s) -
Chaofeng Li,
Yifan Li,
Yunhao Yuan,
Xiaojun Wu,
Qingbing Sang
Publication year - 2018
Publication title -
plos one
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.99
H-Index - 332
ISSN - 1932-6203
DOI - 10.1371/journal.pone.0199430
Subject(s) - distortion (music) , metric (unit) , artificial intelligence , weighting , computer science , image quality , similarity (geometry) , complex wavelet transform , wavelet , wavelet transform , computer vision , image (mathematics) , pattern recognition (psychology) , quaternion , human visual system model , mathematics , discrete wavelet transform , engineering , telecommunications , operations management , geometry , radiology , medicine , amplifier , bandwidth (computing)
Most of real-world image distortions are multiply distortion rather than single distortion. To address this issue, in this paper we propose a quaternion wavelet transform (QWT) based full reference image quality assessment (FR IQA) metric for multiply distorted images, which jointly considers the local similarity of phase and magnitude of each subband via QWT. Firstly, the reference images and distorted images are decomposed by QWT, and then the similarity of amplitude and phase are calculated on each subband, thirdly the IQA metric is constructed by the weighting method considering human visual system (HVS) characteristics, and lastly the scores of each subband are averaged to get the quality score of test image. Experimental results show that the proposed method outperforms the state of art in multiply distorted IQA.