摘要
针对真实失真图像提出一种基于联合字典的无参考(NR)图像质量评价(IQA)方法,分为训练和测试两个阶段。在训练阶段,首先对真实失真图像提取美学特征和自然场景统计特征,然后对图像特征和标签进行联合字典学习,训练得到特征字典和质量字典。在测试阶段,根据特征字典和质量字典计算真实失真图像的质量值。在LIVE Challenge数据库上的实验结果表明,本文方法的评价结果与主观评价结果有较好的相关性,符合人类视觉系统的感知,相比较传统的无参考方法,具有更好的优越性。
In this paper,a no-reference image quality assessment method based on joint dictionary is proposed for authentically distorted images,which can be divided into two stages:training and testing.In the training stage,firstly,the aesthetic features and natural scene statistical features are extracted respectively from the distorted images,and the image features and labels are used to train the feature and quality dictionaries.In the testing stage,the quality of a testing distorted image is calculated according to the feature and quality dictionaries.Experimental results on the LIVE Challenge database show that compared with the traditional non-reference methods,the evaluation results of this method are well correlated with the subjective evaluation results,which accords with the perception of the human visual system.
引文
[1]Ferzli R,Karam L J.A no-reference objective image sharpness metric based on the notion of just noticeable blur(JNB)[J].IEEE Transactions on Image Processing,2009,18(4):717-28.
[2]Saad M A,Bovik A C,Charrier C.Blind image quality assessment:A natural scene statistics approach in the DCT domain[J].IEEE Transactions on Image Processing,2012,21(8):3339-3352.
[3]LI Ke-meng,SHAO Feng,JIANG Qiu-ping,et al.No-reference stereoscopic image quality assessment based on binocular feature combination[J].Journal of Optoelectronics·Laser,2015,26(11):2224-2230.李柯蒙,邵枫,姜求平,等.基于双目特征联合的无参考立体图像质量评价[J].光电子·激光,2015,26(11):2224-2230.
[4]Li L H,Tao D H,Gao X B,et al.Sparse representation for blind image quality assessment[A].Proc.of IEEE Conference on Computer Vision and Pattern Recognition(CVPR)[C].2012,1146-1153.
[5]LI Jun-feng,ZHANG Zhi-xiang,SHEN Jun-min.No-reference image quality assessment based on luminance statistics[J].Journal of Optoelectronics·Laser,2016,25(12):2407-2416.李俊峰,张之祥,沈军民.基于亮度统计的无参考图像质量评价[J].光电子·激光,2016,25(12):2407-2416.
[6]Ghadiyaram D,Bovik A C.Perceptual quality prediction on authentically distorted images using a bag of features approach[J].Journal of Vision,2017,17(1).
[7]Jiang Z,Lin Z,Davis L S.Learning a discriminative dictionary for sparse coding via label consistent K-SVD[A].Proc.of IEEE Conference on Computer Vision and Pattern Recognition[C].2011,1697-1704.
[8]He K,Sun J,Tang X.Single image haze removal using dark channel prior[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2011,33(12):2341-2353.
[9]Luo W,Wang X,Tang X.Content-based photo quality assessment[C].International Conference on Computer Vision.IEEE,2011,2206-2213.
[10]Vu C T,Phan T D,Chandler D M.S3:A spectral and spatial measure of local perceived sharpness in natural images[J].IEEE Transactions on Image Processing,2012,21(3):934-945.
[11]Vos J J.Colorimetric and photometric properties of a 2°fundamental observer[J].Color Research&Application,2010,3(3):125-128.
[12]Datta R,Joshi D,Li J,et al.Studying aesthetics in photographic images using a computational approach[A].Proc.of European Conference on Computer Vision[C].2006,288-301.
[13]Cohen-Or D,Sorkine O,Ran G,et al.Color harmonization[A].ACM Transactions on Graphics,2006,25(3):624-630.
[14]Mittal A,Moorthy A K,Bovik A C.No-reference image quality assessment in the spatial domain[J].IEEE Transactions on Image Processing,2012,21(12):4695-4708.
[15]Xue W,Mou X,Zhang L,et al.Blind image quality assessment using joint statistics of gradient magnitude and Laplacian features[J].IEEE Transactions on Image Processing,2014,23(11):4850-4862.
[16]Moorthy A K,Bovik A C.Blind image quality assessment:From natural scene statistics to perceptual quality[J].IEEE Transactions on Image Processing,2011,20(12):3350-3364.
[17]Aharon M,Elad M,Bruckstein A,et al.K-SVD:An algorithm for designing overcomplete dictionaries for sparse representation[J].IEEE Transactions on Signal Processing,2006,54(11):4311-4322.
[18]Pat Yi,Rezaiifar R,Krishnaprasad P,et al.Orthogonal matching pursuit:recursive function approximation with applications to wavelet decomposition[A].Proc.of Asilimar Signals,Systems and Computers[C].1993,40-44.
[19]Xue W,Zhang L,Mou X.Learning without human scores for blind image quality assessment[A].Proc.of IEEE Conference on Computer Vision and Pattern Recognition[C].2013,995-1002.
[20]Mittal A,Soundararajan R,Bovik A C.Making a“completely blind”image quality analyzer[J].IEEE Signal Processing Letters,2013,20(3):209-212.
[21]Zhang L,Zhang L,Bovik A C.A feature-enriched completely blind image qaulity evaluator[J].IEEE Transactions on Image Processing,2015,24(8):2579-2591.
[22]Moorthy A K,Bovik A C.A two-step framework for constructing blind image quality indices[J].IEEE Signal Processing Letters,2010,17(5):513-516.