曾台英,邵雪,汪祖辉.基于局部离散小波变换的无参考模糊图像质量评价方法[J].光电子激光,2016,27(7):754~760 |
基于局部离散小波变换的无参考模糊图像质量评价方法 |
No-reference blur image quality assessment based on local discrete wavelet tran sform |
投稿时间:2015-12-21 |
DOI: |
中文关键词: 无参考(NR)图像质量评价 局部离散小波变换 模糊估计 人类视觉 |
英文关键词:no-reference image quality assessment local discrete wavelet transf orm blur estimation human vision |
基金项目:国家重大科学仪器设备开发专项(2011YQ150040)和新闻出版总 署数字印刷工程研究中心数字传播重点实验室基金(10-00-309-000)资助项目 (上海理工大学 出版印刷与艺术设计学院,上海 200093) |
|
摘要点击次数: 905 |
全文下载次数: 0 |
中文摘要: |
构造模糊图像的参考图像,使无参考(NR)图像质 量评价转变为一种全参考(FR)图像质量评价,提出一种基于 局部离散小波变换的NR模糊图像质量评价方法。首先,采用低通滤波的方 法构造原始模糊图 像的参考图像;然后,通过局部离散小波变换提取出原始模糊图像和参考图像的高频能量图 ;最后计 算两幅高频能量图的相似度,得到原始模糊图像的质量。在LIVE、TID2008和CSIQ 3个数据 库上的实验结果表明, 本文算法的评 价性能较优,相较其他常用的4个清晰度评价算法,在与主观评价的单调性和一致性上有明 显提高,在LIVE 数据库上的SROCC值达到0.9598,在CISQ数 据库上的SROCC值达到0.9104 ,与主观评价平均意见成绩(MOS)值拟合的散点图拟合度为最优。实验数据 结果表明,本文算法性能稳定,符合人类视觉系统。 |
英文摘要: |
Construct the reference image for blur images,so that no reference i mage quality assessment is transformed into a full reference image quality evaluation,a no-reference blur image quality assessment based on local discrete wavelet transform is proposed.Firstly,the reference ima ge of the original blur image is constructed by the method of low pass filtering;then,the high frequency energy graphs of the original fuzzy image and reference image are extracted by local discrete wavelet transform;finally,the s imilarity of two high frequency energy graphs is calculated,and the quality of the original blur image is obtain ed.Th e experimental results on 3databases show that the proposed algorithm is superior to the other four definition evalua tion algorithms,and has a significant improvement on the monotonic and consistency with subjective evaluat ion.The spearman rank-order coorelation coefficient (SROCC) value reaches 0.9598in LIVE database.The performance of the algorithm is s table,and the algorithm is more consistent with the human visual system. |
查看全文 下载PDF阅读器 |
关闭 |
|
|
|
|
|