• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2011, Vol. 33 ›› Issue (2): 133-136.doi: 10.3969/j.issn.1007130X.2011.

• 论文 • 上一篇    下一篇

基于边缘的结构相似度模糊图像质量评价

戚尚菊,纪秀花   

  1. (山东经济学院计算机科学与技术学院,山东 济南 250014) 
  • 收稿日期:2010-04-01 修回日期:2010-06-25 出版日期:2011-02-25 发布日期:2011-02-25
  • 通讯作者: 戚尚菊
  • 作者简介:戚尚菊(1964),女,山东新泰人,硕士,副教授,研究方向为数字图像处理。纪秀花(1964),女,山东平原人,博士生,教授,研究方向为数字图像处理。
  • 基金资助:

    国家自然科学基金资助项目(60573114);山东省教育厅科技计划项目(J07YJ10)

lurred Image Quality Assessment Based on Edge Structural Similarity

QI Shangju,JI Xiuhua   

  1. (School of Computer Science and Technology,Shandong Economic University,Jinan 250014,China)
  • Received:2010-04-01 Revised:2010-06-25 Online:2011-02-25 Published:2011-02-25

摘要:

结构相似性理论是一种关于图像质量评价的新思想,它很好地模拟了人眼视觉特性的整体功能。本文首先介绍了传统图像质量评价方法的不足,分析了基于视觉特性的结构相似度理论。作为结构相似性理论的一个实现,结构相似度模型(SSIM)简单且评价性能优于峰值信噪比(PSNR)或均方误差(MSE),但SSIM模型不能较好地评价严重模糊的降质图像。基于此,在SSIM基础上,本文提出了一种新颖的、基于边缘的图像质量评价模型(ESSIM)。该模型充分考虑了图像的边缘信息和人类视觉系统的关系。实验结果表明,ESSIM是一种有效的图像质量评价方法,尤其在对模糊图像的质量评价上优于结构相似度的评价方法SSIM。

关键词: 图像质量评价, 结构相似度, 图像边缘, 人眼视觉特性

Abstract:

The philosophy of structural similarity is a new idea about image quality assessment, which models the low level composition of Human Visual Systems(HVS). The paper introduces the shortcomings of the traditional image quality assessment methods firstly,and analyzes the structural similarity theory based the human visual systems. As an implementation of the new philosophy, the Structural Similarity model(SSIM) is simple and has been proved to be better than the PSNR (Peak Signal to Noise Ratio) or the MSE (Mean Square Error)model, but there still remain some deficiencies in assessing badly blurred images. On the foundation of SSIM,the paper proposes a new image quality assessment method based on the Edge Structural Similarity(ESSIM). It makes use of the relationship between the edge of image and the human visual system. The experimental results show that the ESSIM model is an effective method, and more consistent than the SSIM model.

Key words: image quality assessment;structural similarity;image edge;human visual system