基于双向过滤法的块效应下无参考图像质量评价  

No-Reference Image Quality Assessment Under Blocking Artifacts Based on Bidirectional Filtration

作　　者：马璐[1,2] 杨文飞 MA Lu;YANG Wenfei(Department of Information and Computer,Suzhou Vocational and Technical College,Suzhou 234000,Anhui,China;School of Electronic Science and Engineering,Nanjing University,Nanjing 210023,Jiangsu,China;School of Foreign Languages,Suzhou University,Suzhou 234000,Anhui,China)

机构地区：[1]宿州职业技术学院计算机信息系,安徽宿州234000 [2]南京大学电子科学与工程学院,江苏南京210023 [3]宿州学院外国语学院,安徽宿州234000

出　　处：《江汉大学学报（自然科学版）》2020年第1期91-96,共6页Journal of Jianghan University：Natural Science Edition

基　　金：安徽省教育厅高等学校自然科学重点项目(KJ2017A657);安徽省教育厅高校优秀青年骨干人才项目(gxgnfx2018112)

摘　　要：针对块效应下无参考图像质量评价,提出了双向过滤法(bidirectional filtration),通过分辨图像真边界与假边界进行块效应下图像评价。该方法从两个不同的场域对图像边界进行逐层过滤,达到对真边界的有效定位,最后通过评价模型得出评价值。实验测试结果得出的数据与SSIM模型相比有了一定的改进。To evaluate the image quality without reference under blocking effect,the author proposed a bidirectional filtration(BIF) method,which aims at providing a new idea for boundary selection,and resolved the image evaluation under blocking effect more effectively by distinguishing the true and false boundary of the image.The method filtered the image boundary layer by layer from two different fields to locate the true boundary effectively.Finally,the evaluation values were obtained through the evaluation model.Compared with SSIM model,the data obtained from the experimental results were improved to a certain extent.It is concluded that this method is more accurate in evaluating the no-reference image under blocking effect.

关 键 词：双向过滤法 图像质量评价 图像边界 

分 类 号：TN911.73[电子电信—通信与信息系统]