·1004· 工程科学学报.第40卷，第8期 由于BLIINDS-I对图像进行离散余弦变换 2012,21(12):4695 (discrete cosine transform,DCT),因此有很高的计算 [9]Hubel D H,Wiesel T N.Receptive fields,binocular interaction 复杂度.本文算法由于计算条件嫡需要多次遍历图 and functional architecture in the cat's visual cortex.Physiol, 1962,160(1):106 像，因此运行时间略慢OG_IQA和BRISQUE算法. [10]Clark M.Bovik A C.Experiments in segmenting texton patterns 总体而言，可以认为该算法有较低的复杂度 using localized spatial filters.Pattern Recognit,1989,22(6): 707 4结论 [11]Marziliano P,Dufaux F.Winkler S,et al.A no-reference per- (1)本文提出了一种新的无参考质量评价方 ceptual blur metric /IEEE International Conference on Image 法，为了避免单一算子的局限性，同时考虑一阶边缘 Processing (ICIP).Rochester,2002:Ill-57 [12]Marziliano P.Dufaux F,Winkler S,et al.Perceptual blur and 特征GM和二阶边缘特征LOG,并利用它们之间的 ringing metrics:application to JPEG2000.Signal Process:Image 相似度来预测图像质量. Commun,2004,19(2):163 (2)考虑到了全局特征与局部特征的关系，同 [13]Liu L X,Hua Y,Zhao Q J,et al.Blind image quality assess- 时使用全局的GM特征与改进的局部RM特征，能 ment by relative gradient statistics and adaboosting neural net- 够更好的反映出失真对图像纹理细节的破坏. work.Signal Process:Image Commun,2016,40:1 (3)将信息论中的条件嫡引入来衡量两种特征 [14]Zhang M,Muramatsu C,Zhou X R,et al.Blind image quality assessment using the joint statistics of generalized local binary 的相似性，不仅可以很好的表达出图像的失真程度， pattern.IEEE Signal Process Lett,2014,22(2):207 而且大大减少了特征的维数，本算法仅使用10维特 [15]Li Q H,Lin W S,Fang Y M.No-reference quality assessment 征就能达到理想的效果 for multiply-distorted images in gradient domain.IEEE Signal (4)本文算法具有很高的主观一致性，符合人 Process Lett,2016,23(4):541 眼的视觉系统，统计结果表明该算法性能优于大部 [16]Yue G H,Hou C P,Gu K,et al.No reference image blurriness assessment with local binary patterns.J Visual Commun Image 分主流的无参考质量评价算法，在不同训练-测试 Representation,2017,49:382 比例下均有着较好的性能，具有高度的鲁棒性，并且 [17]Sheikh H R,Wang Z,Cormack L,et al.LIVE image quality as- 有较低的时间复杂度，可以满足实时性需要. sessment database release 2 J/OL].Laboratory for Image Video Engineering (2005 [2017-08-22 ]http://live.ece. 参考文献 utexas.edu/research/quality [1]Wang Q.Gu K,Zhang X,et al.Subjective and objective quali- [18]Ponomarenko N,Lukin V,Zelensky A,et al.TID2008-a data- ty assessment of compressed sereen content images.IEEE J Emer. base for evaluation of full-reference visual quality assessment met- ging Sel Top Circuits Syst,2016,6(4):532 rics.Adr Mod Radioelectron,2009,10(4):30 [2]Zhang X F,Wang S Q,Gu K,et al.Just-noticeable difference- [19]Ghosh K,Sarkar S,Bhaumik K.Understanding image structure based perceptual optimization for JPEG compression.IEEE Signal from a new multi-scale representation of higher order derivative Process Lett,2017,24(1):96 filters.Image Vision Comput,2007,25(8):1228 [3]Li L D,Yan Y,Lu Z L,et al.No-reference quality assessment of [20]Gu K,Li L D,Lu H,et al.A fast reliable image quality predic- deblurred images based on natural scene statistics.IEEE Access, tor by fusing micro-and macro-structures.IEEE Trans Ind Elec- 2017,5:2163 tron,2017,64(5):3903 [4]Gu K,Zhai G T,Wang S Q,et al.A general histogram modifica- [21]Wang Z,Bovik A C,Sheikh H R,et al.Image quality assess- tion framework for efficient contrast enhancement /IEEE Interna- ment:from error visibility to structural similarity.IEEE Trans tional Symposium on Circuits and Systems.Lisbon,2015:2816 Image Process,2004,13(4):600 [5]Ruderman DL.The statistics of natural images.Network Comput [22]Ye P,Kumar J.Kang L,et al.Unsupervised feature leaming Neural Syst,1994,5(4):517 framework for no-reference image quality assessment //EEE [6]Saad M A,Bovik A C,Charrier C.Blind image quality assess- Conference on Computer Vision and Pattern Recognition(CVPR). ment:a natural scene statistics approach in the DCT domain. Providence,2012:1098 IEEE Trans Image Process,2012,21(8):3339 [23]Zhang M,Muramatsu C.Zhou X R,et al.Blind image quality [7]Moorthy A K,Bovik A C.Blind image quality assessment:from assessment using the joint statistics of generalized local binary natural scene statistics to perceptual quality.IEEE Trans Image pattern.IEEE Signal Process Lett,2014,22(2):207 Process,2011,20(12):3350 [24]Gu K,Zhai G T,Yang X K,et al.Using free energy principle [8]Mittal A,Moorthy A K.Bovik A C.No-reference image quality for blind image quality assessment.IEEE Trans Multimedia, assessment in the spatial domain.IEEE Trans Image Process, 2015,17(1):50工程科学学报,第 40 卷,第 8 期 由于 BLIINDS鄄II 对图像进行离散余弦变换 (discrete cosine transform, DCT),因此有很高的计算 复杂度. 本文算法由于计算条件熵需要多次遍历图 像,因此运行时间略慢 OG_IQA 和 BRISQUE 算法. 总体而言,可以认为该算法有较低的复杂度. 4 结论 (1) 本文提出了一种新的无参考质量评价方 法,为了避免单一算子的局限性,同时考虑一阶边缘 特征 GM 和二阶边缘特征 LOG,并利用它们之间的 相似度来预测图像质量. (2) 考虑到了全局特征与局部特征的关系,同 时使用全局的 GM 特征与改进的局部 RM 特征,能 够更好的反映出失真对图像纹理细节的破坏. (3) 将信息论中的条件熵引入来衡量两种特征 的相似性,不仅可以很好的表达出图像的失真程度, 而且大大减少了特征的维数,本算法仅使用 10 维特 征就能达到理想的效果. (4) 本文算法具有很高的主观一致性,符合人 眼的视觉系统,统计结果表明该算法性能优于大部 分主流的无参考质量评价算法,在不同训练鄄鄄 测试 比例下均有着较好的性能,具有高度的鲁棒性,并且 有较低的时间复杂度,可以满足实时性需要. 参 考 文 献 [1] Wang S Q, Gu K, Zhang X, et al. Subjective and objective quali鄄 ty assessment of compressed screen content images. IEEE J Emer鄄 ging Sel Top Circuits Syst, 2016, 6(4): 532 [2] Zhang X F, Wang S Q, Gu K, et al. Just鄄noticeable difference鄄 based perceptual optimization for JPEG compression. IEEE Signal Process Lett, 2017, 24(1): 96 [3] Li L D, Yan Y, Lu Z L, et al. No鄄reference quality assessment of deblurred images based on natural scene statistics. IEEE Access, 2017, 5: 2163 [4] Gu K, Zhai G T, Wang S Q, et al. A general histogram modifica鄄 tion framework for efficient contrast enhancement / / IEEE Interna鄄 tional Symposium on Circuits and Systems. Lisbon, 2015: 2816 [5] Ruderman D L. The statistics of natural images. Network Comput Neural Syst, 1994, 5(4): 517 [6] Saad M A, Bovik A C, Charrier C. Blind image quality assess鄄 ment: a natural scene statistics approach in the DCT domain. IEEE Trans Image Process, 2012, 21(8): 3339 [7] Moorthy A K, Bovik A C. Blind image quality assessment: from natural scene statistics to perceptual quality. IEEE Trans Image Process, 2011, 20(12): 3350 [8] Mittal A, Moorthy A K, Bovik A C. No鄄reference image quality assessment in the spatial domain. IEEE Trans Image Process, 2012, 21(12): 4695 [9] Hubel D H, Wiesel T N. Receptive fields, binocular interaction and functional architecture in the cat蒺s visual cortex. J Physiol, 1962, 160(1): 106 [10] Clark M, Bovik A C. Experiments in segmenting texton patterns using localized spatial filters. Pattern Recognit, 1989, 22 (6): 707 [11] Marziliano P, Dufaux F, Winkler S, et al. A no鄄reference per鄄 ceptual blur metric / / IEEE International Conference on Image Processing (ICIP). Rochester, 2002: III鄄57 [12] Marziliano P, Dufaux F, Winkler S, et al. Perceptual blur and ringing metrics: application to JPEG2000. Signal Process: Image Commun, 2004, 19(2): 163 [13] Liu L X, Hua Y, Zhao Q J, et al. Blind image quality assess鄄 ment by relative gradient statistics and adaboosting neural net鄄 work. Signal Process: Image Commun, 2016, 40: 1 [14] Zhang M, Muramatsu C, Zhou X R, et al. Blind image quality assessment using the joint statistics of generalized local binary pattern. IEEE Signal Process Lett, 2014, 22(2): 207 [15] Li Q H, Lin W S, Fang Y M. No鄄reference quality assessment for multiply鄄distorted images in gradient domain. IEEE Signal Process Lett, 2016, 23(4): 541 [16] Yue G H, Hou C P, Gu K, et al. No reference image blurriness assessment with local binary patterns. J Visual Commun Image Representation, 2017, 49: 382 [17] Sheikh H R, Wang Z, Cormack L, et al. LIVE image quality as鄄 sessment database release 2 [ J/ OL]. Laboratory for Image & Video Engineering (2005) [2017鄄鄄 08鄄鄄 22 ]. http: / / live. ece. utexas. edu / research / quality [18] Ponomarenko N, Lukin V, Zelensky A, et al. TID2008鄄鄄 a data鄄 base for evaluation of full鄄reference visual quality assessment met鄄 rics. Adv Mod Radioelectron, 2009, 10(4): 30 [19] Ghosh K, Sarkar S, Bhaumik K. Understanding image structure from a new multi鄄scale representation of higher order derivative filters. Image Vision Comput, 2007, 25(8): 1228 [20] Gu K, Li L D, Lu H, et al. A fast reliable image quality predic鄄 tor by fusing micro鄄 and macro鄄structures. IEEE Trans Ind Elec鄄 tron, 2017, 64(5): 3903 [21] Wang Z, Bovik A C, Sheikh H R, et al. Image quality assess鄄 ment: from error visibility to structural similarity. IEEE Trans Image Process, 2004, 13(4): 600 [22] Ye P, Kumar J, Kang L, et al. Unsupervised feature learning framework for no鄄reference image quality assessment / / IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Providence, 2012: 1098 [23] Zhang M, Muramatsu C, Zhou X R, et al. Blind image quality assessment using the joint statistics of generalized local binary pattern. IEEE Signal Process Lett, 2014, 22(2):207 [24] Gu K, Zhai G T, Yang X K, et al. Using free energy principle for blind image quality assessment. IEEE Trans Multimedia, 2015, 17(1): 50 ·1004·