电子学报 ›› 2014, Vol. 42 ›› Issue (7): 1419-1423.DOI: 10.3969/j.issn.0372-2112.2014.07.027

• 科研通信 • 上一篇    下一篇

一个图像质量盲评估的统计测度

黄虹, 张建秋   

  1. 复旦大学电子工程系, 上海 200433
  • 收稿日期:2013-03-22 修回日期:2013-12-08 出版日期:2014-07-25
    • 作者简介:
    • 黄虹 女.1988年8月出生,福建宁德人.2010年本科、2013年硕士毕业于复旦大学电子工程系,在校期间从事数字信号处理方面的有关研究.E-mail:10210720031@fudan.edu.cn;张建秋 男.1962年生于湖南省隆回县.现任复旦大学电子工程系教授、博士生导师,IEEE高级会员,主要研究领域有信息处理理论及其在测量和仪器、新型传感器、控制和通信中的应用.E-mail:jqzhang01@fudan.edu.cn
    • 基金资助:
    • 国家自然科学基金 (No.61171127)

A Statistical Measure for Blind Image Quality Assessment

HUANG Hong, ZHANG Jian-qiu   

  1. Department of Electronic Engineering, Fudan University, Shanghai 200433, China
  • Received:2013-03-22 Revised:2013-12-08 Online:2014-07-25 Published:2014-07-25

摘要:

本文提出了一个图像质量盲评估的统计测度.该测度首先根据自然图像的统计性质与失真图像的模型,实现对图像小波系数分布参数的盲估计;再利用估计的分布参数来计算失真图像与参考图像之间的互信息,以量化失真图像对参考图像的保真度,进而实现对图像质量的评估.本文提出的测度避免了对参考图像的依赖,且克服了现有图像质量盲评估对特征选择与提取、机器学习等过程的依赖.LIVE图像质量评估数据库的总体评估结果表明:本文提出的盲评估统计测度对图像质量评估结果与数据库的主观评估结果高度一致,且优于文献中报道的盲评估测度.

关键词: 图像质量盲评估, 保真度, 互信息, 自然场景统计模型, 高斯尺度混合模型

Abstract:

A statistical measure for blind image quality assessment (IQA) is proposed.The wavelet coefficients' distribution parameters of the distorted images are blind estimated based on the natural scene statistics and the image distortion model;the mutual information between the distorted and the corresponding reference images is further calculated from the estimated distribution parameters.The quantified information fidelity is regarded as an efficient image quality assessment criterion.The proposed statistical measure in this paper does not require any prior information of the reference image and avoids the feature selection,feature extraction and machine learning processes required by existing blind image quality assessment methods.Evaluated on the LIVE IQA database,it is demonstrated that the proposed statistical measure corresponds well with the subjective human evaluations and outperforms the state-of-art blind IQA algorithms.

Key words: blind image quality assessment, mutual information, information fidelity, natural scene statistics (NSS) model, Gaussian scale mixture (GSM) model

中图分类号: