中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 国家天文台
  3. 中国科学院国家天文台
  4. 太阳物理研究部
Multi-Task Rank Learning for Image Quality Assessment

文献类型：期刊论文

作者Yan, Yihua2; Xu, Long2; Li, Jia3,4; Lin, Weisi5; Zhang, Yongbing6; Ma, Lin7; Fang, Yuming1
刊名IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
出版日期2017-09-01
卷号27期号:9页码:1833-1843
关键词Image Quality Assessment (Iqa) Machine Learning (Ml) Mean Opinion Score (Mos) Pairwise Comparison Rank Learning
DOI10.1109/TCSVT.2016.2543099
文献子类Article
英文摘要In practice, images are distorted by more than one distortion. For image quality assessment (IQA), existing machine learning (ML)-based methods generally establish a unified model for all the distortion types, or each model is trained independently for each distortion type, which is therefore distortion aware. In distortion-aware methods, the common features among different distortions are not exploited. In addition, there are fewer training samples for each model training task, which may result in overfitting. To address these problems, we propose a multi-task learning framework to train multiple IQA models together, where each model is for each distortion type; however, all the training samples are associated with each model training task. Thus, the common features among different distortion types and the said underlying relatedness among all the learning tasks are exploited, which would benefit the generalization ability of trained models and prevent overfitting possibly. In addition, pairwise image quality ranking instead of image quality rating is optimized in our learning task, which is fundamentally departed from traditional ML-based IQA methods toward better performance. The experimental results confirm that the proposed multi-task rank-learning-based IQA metric is prominent against all state-of-the-art nonreference IQA approaches.
WOS关键词VISUAL SALIENCY ESTIMATION ; NATURAL SCENE STATISTICS ; ALGORITHMS ; REGRESSION ; FRAMEWORK ; JPEG2000 ; DOMAIN
WOS研究方向Engineering
语种英语
WOS记录号WOS:000409531400001
资助机构National Natural Science Foundation (NSFC) of China(61202242 ; National Natural Science Foundation (NSFC) of China(61202242 ; CAS ; CAS ; National Natural Science Foundation of China(61370113 ; National Natural Science Foundation of China(61370113 ; 61572461) ; 61572461) ; U1201255 ; U1201255 ; U1301257 ; U1301257 ; 61571212 ; 61571212 ; 11433006) ; 11433006) ; National Natural Science Foundation (NSFC) of China(61202242 ; National Natural Science Foundation (NSFC) of China(61202242 ; CAS ; CAS ; National Natural Science Foundation of China(61370113 ; National Natural Science Foundation of China(61370113 ; 61572461) ; 61572461) ; U1201255 ; U1201255 ; U1301257 ; U1301257 ; 61571212 ; 61571212 ; 11433006) ; 11433006)
源URL[http://ir.bao.ac.cn/handle/114a11/20087]  
专题国家天文台_太阳物理研究部
作者单位1.Jiangxi Univ Finance & Econ, Sch Informat Technol, Nanchang 330032, Jiangxi, Peoples R China
2.Chinese Acad Sci, Natl Astron Observ, Key Lab Solar Act, Beijing 100012, Peoples R China
3.Beihang Univ, Sch Comp Sci & Engn, State Key Lab Virtual Real Technol & Syst, Beijing 100191, Peoples R China
4.Beihang Univ, Int Res Inst Multidisciplinary Sci, Beijing 100191, Peoples R China
5.Nanyang Technol Univ, Dept Comp Engn, Singapore 639798, Singapore
6.Tsinghua Univ, Grad Sch Shenzhen, Shenzhen 518005, Peoples R China
7.Huawei Noahs Ark Lab, Hong Kong, Hong Kong, Peoples R China
推荐引用方式
GB/T 7714		 Yan, Yihua,Xu, Long,Li, Jia,et al. Multi-Task Rank Learning for Image Quality Assessment[J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,2017,27(9):1833-1843.
APA Yan, Yihua.,Xu, Long.,Li, Jia.,Lin, Weisi.,Zhang, Yongbing.,...&Fang, Yuming.(2017).Multi-Task Rank Learning for Image Quality Assessment.IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,27(9),1833-1843.
MLA Yan, Yihua,et al."Multi-Task Rank Learning for Image Quality Assessment".IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY 27.9(2017):1833-1843.

入库方式： OAI收割

来源：国家天文台

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。