摘 要：针对视觉显著（visual saliency index，VSI）图像质量评价的不足，通过引入人眼的感知特性，提出了一种视觉特征融合（visual saliency feature pooling，VSFP）的评价方法。VSFP 方法首先对失真图像的灰度特征进行评价，作为能量补充信息，然后基于人眼的中央凹生理特性对图像局部特征评价进行加权融合，最后基于回归方程对多特征评价进行自适应融合。实验表明所提方法明显提高了VSI 方法的评价性能。
关键词：图像质量评价；特征融合；视觉中央凹；回归方程
中图分类号：TP391.41 文献标识码：A 文章编号：2096-4706（2019）07-0080-03
Image Assessment Based on Feature Pooling of Visual Saliency
WANG Saijiao1，2
（1.Taizhou Radio and Television University，Taizhou 318000，China；2.Computer College，Hangzhou Dianzi University，Hangzhou 310018，China）
Abstract：Aiming at the shortcomings of visual saliency index（VSI）for image quality assessment，an assessment method called visual saliency feature pooling（VSFP）is proposed by introducing the perceptual characteristics of human eyes. Firstly，the VSFP method evaluates the gray level features of distorted images are assessed and viewed as energy supplement information. Secondly， the local features assessment is pooled with the weights based on the physiological characteristics of the central foveal of the human eye. Finally，the multi-features assessment is adaptively pooled based on the regression equation. Experiments show that the assessment performance of the proposed method is significantly improved compared to VSI method.
Keywords：image quality assessment；feature pooling；visual foveal；regression equation
参考文献：
[1] 王勇，王宇庆，赵晓晖. 图像质量客观评价的复数矩阵结构相似度方法 [J]. 仪器仪表学报，2014，35（5）：1118-1129.
[2] Guo Yingchun，Hao Yuting，Yu Ming.Image retargeting quality assessment based on content deformation measurement [J].Signal Processing：Image Communication，2018，67：171-181.
[3] Wang Z，Bovik AC，Sheikh HR，et al.Image Quality Assessment：From Error Visibility to Structural Similarity [J].IEEE Transactions on Image Processing，2004，13（4）：600-612.
[4] Zhang L，Shen Y，Li H.VSI：A Visual Saliency-Induced Index for Perceptual Image Quality Assessment [J].IEEE Transactions on Image Processing，2014，23（10）：4270-4281.
[ 5 ] L i u A n m i n，L i n W e i s i，N a r w a r i a M . I m a g e q u a l i t y assessment based on gradient similarity [J].IEEE Transactions on Image Processing，2012，21（4）：1500-1512.
作者简介：王赛娇（1977-），女，汉族，浙江人，讲师，硕士， 研究方向：计算机视觉，人工智能。