数字视频监控实时降噪算法研究
|
摘要
视频监控具有直观、方便、可靠、信息内容丰富等优点，因此被广泛应用于金融、商业、交通、住宅、社区等场合，为这些领域的环境监控和安全防范起到了不可忽略的作用。然而，噪声的存在一直是实时视频监控中不可回避的问题，也成为影响视频视觉质量的重要因素，因此，降噪技术是增强视频视觉质量的一种非常重要的手段。
视频图像降噪算法中，在保护图像边缘细节和提高峰值信噪比方面，时域滤波要优于空域滤波，但需要引入运动估计来更好地利用视频帧在时域上的相关性，而噪声的存在容易影响运动估计的准确度，从而降低整个算法的降噪性能。为了使降噪算法适合多种噪声级别，提高运动估计的准确度，需要在现有的降噪算法中引入自适应机制，这方面有待探索。同时，目前基于运动估计的视频降噪算法受运算量的限制较大，因而限制了其在视频实时监控系统中的应用。因此，迫切需要寻找一种适合多种噪声级别的快速视频降噪算法。另外，视频图像的降噪算法和运动检测阈值的设定都不同程度地依赖于噪声大小的先验，但目前没有一种噪声估计方法在高质量、高噪声、图像含有大量纹理信息及剧烈运动物体的情况下，都能给出相当准确的结果。因此，研究精确稳健的视频噪声估计算法是非常必要的。
本文从上述需求出发，主要工作如下：1、提出了一种基于块内邻域相关度的差分视频噪声估计算法，此算法充分利用了视频信号时域上的相关性，在高质量、高噪声、图像含有大量纹理信息及含有剧烈运动物体的情况下都获得了相当准确的估计。2、在基于时空联合的实时视频降噪算法基础上研究了自适应运动检测阈值设定方法，使得运动检测具有自适应性且更加准确。3、提出了一种基于块结构相似度的快速运动估计算法，此算法结合了图像结构相似度理论进行运动估计，对运动的判断比较准确，且算法运算速度较快，在降噪算法中能更好地满足视频处理的实时性需求。
Video Monitoring, because of its advantages of intuition, convenience, reliability,rich information, and so on, is widely applied in areas such as finance, commerce,traffic, house, and community, and plays an important role in monitoring andprotection of these areas. However, the existence of the noise makes it an inevitableproblem in real-time video monitoring, and an important factor to evaluate the qualityof video vision.
Among the video denoising algorithms, the time domain filtering algorithm issuperior than spatial filtering algorithm in the protection of frame edge and improvingPSNR. However, the motion evaluation is necessary to take advantage of correlationof video frames in time domain, although, due to the existence of the noise, theaccuracy of the motion evaluation is influenced, the performance of the algorithm isinfluenced and then. To improve the accuracy and achieve compatibility amongnoises of many levels, an adaptive algorithm is essential, on which the research isimperative. At present, because of the high computation load, the application of thede-noising algorithms in real-time video monitoring is limited. For that, a time-savingvideo denoising algorithms for noises of many levels is eagerly needed. Besides, thespatial filtering algorithm and the threshold setting of motion detection is more or lessrely on prior knowledge of the noises, however, it’s a pity that none of the exist noiseestimation algorithms can achieve perfect results in high quality, high noise, largeamount of texture information and severely motion conditions. So, a steady noiseestimation algorithm with a higher accuracy is essential.
In this paper, three efforts are involved:
1. A differential video noise estimation algorithm based on block neighborhoodcorrelation is demonstrated, which takes full advantages of the video signalcorrelation in time domain, can achieve results with higher accuracy in highquality, high noise, large amount of texture information and severely motionconditions.
2. A threshold setting algorithm for adaptive motion detection is demonstratedwhich improves the algorithm of video real-time denoising based on spatio-temporal combination, and leads to a higher accuracy and an adaptive performance.
3. A fast motion estimation algorithm based on block structure similarity is demonstrated, which combines image structure similarity theory. With this algorithm, the time saving desire is realized, a higher accuracy is achieved,and the real-time demand is satisfied.
视频图像降噪算法中，在保护图像边缘细节和提高峰值信噪比方面，时域滤波要优于空域滤波，但需要引入运动估计来更好地利用视频帧在时域上的相关性，而噪声的存在容易影响运动估计的准确度，从而降低整个算法的降噪性能。为了使降噪算法适合多种噪声级别，提高运动估计的准确度，需要在现有的降噪算法中引入自适应机制，这方面有待探索。同时，目前基于运动估计的视频降噪算法受运算量的限制较大，因而限制了其在视频实时监控系统中的应用。因此，迫切需要寻找一种适合多种噪声级别的快速视频降噪算法。另外，视频图像的降噪算法和运动检测阈值的设定都不同程度地依赖于噪声大小的先验，但目前没有一种噪声估计方法在高质量、高噪声、图像含有大量纹理信息及剧烈运动物体的情况下，都能给出相当准确的结果。因此，研究精确稳健的视频噪声估计算法是非常必要的。
本文从上述需求出发，主要工作如下：1、提出了一种基于块内邻域相关度的差分视频噪声估计算法，此算法充分利用了视频信号时域上的相关性，在高质量、高噪声、图像含有大量纹理信息及含有剧烈运动物体的情况下都获得了相当准确的估计。2、在基于时空联合的实时视频降噪算法基础上研究了自适应运动检测阈值设定方法，使得运动检测具有自适应性且更加准确。3、提出了一种基于块结构相似度的快速运动估计算法，此算法结合了图像结构相似度理论进行运动估计，对运动的判断比较准确，且算法运算速度较快，在降噪算法中能更好地满足视频处理的实时性需求。
Video Monitoring, because of its advantages of intuition, convenience, reliability,rich information, and so on, is widely applied in areas such as finance, commerce,traffic, house, and community, and plays an important role in monitoring andprotection of these areas. However, the existence of the noise makes it an inevitableproblem in real-time video monitoring, and an important factor to evaluate the qualityof video vision.
Among the video denoising algorithms, the time domain filtering algorithm issuperior than spatial filtering algorithm in the protection of frame edge and improvingPSNR. However, the motion evaluation is necessary to take advantage of correlationof video frames in time domain, although, due to the existence of the noise, theaccuracy of the motion evaluation is influenced, the performance of the algorithm isinfluenced and then. To improve the accuracy and achieve compatibility amongnoises of many levels, an adaptive algorithm is essential, on which the research isimperative. At present, because of the high computation load, the application of thede-noising algorithms in real-time video monitoring is limited. For that, a time-savingvideo denoising algorithms for noises of many levels is eagerly needed. Besides, thespatial filtering algorithm and the threshold setting of motion detection is more or lessrely on prior knowledge of the noises, however, it’s a pity that none of the exist noiseestimation algorithms can achieve perfect results in high quality, high noise, largeamount of texture information and severely motion conditions. So, a steady noiseestimation algorithm with a higher accuracy is essential.
In this paper, three efforts are involved:
1. A differential video noise estimation algorithm based on block neighborhoodcorrelation is demonstrated, which takes full advantages of the video signalcorrelation in time domain, can achieve results with higher accuracy in highquality, high noise, large amount of texture information and severely motionconditions.
2. A threshold setting algorithm for adaptive motion detection is demonstratedwhich improves the algorithm of video real-time denoising based on spatio-temporal combination, and leads to a higher accuracy and an adaptive performance.
3. A fast motion estimation algorithm based on block structure similarity is demonstrated, which combines image structure similarity theory. With this algorithm, the time saving desire is realized, a higher accuracy is achieved,and the real-time demand is satisfied.
引文
[1] Brailean J.C. Noise reduction filters for dynamic image sequences:a review[J].Proceedings of the IEEE,1995,83(9):1272-1292.
[2] JingYu Yang. Image and video denoising using adaptive dual-tree discretewavelet packets[J].IEEE Transactions on Circuits and Systems for Video Technology,2009,19(5):642-655.
[3] Luisier F. SURE-LET for orthonormal wavelet-domain video denoising[J].IEEE Transactions on Circuits and Systems for Video Technology,2010,20(6):913-919.
[4] Shigong Yu. Video denoising using motion compensated3-D wavelet transformwith integrated recursive temporal filtering[J].IEEE Transactions on Circuits andSystems for Video Technology,2010,20(6):780-791.
[5]孙延奎.小波分析及其应用[M].北京:机械工业出版社,2005.
[6]程正兴,杨守志,冯晓霞.小波分析的理论、算法、进展和应用[M].北京:国防工业出版社,2007.
[7]安世霞.基于3D小波的视频降噪算法研究[D].兰州:兰州大学,2009.
[8]谭洪涛.视频图像降噪关键技术研究[D].重庆:重庆大学,2010.
[9] Volodymyr Ponomaryov,Alberto Rosales-Silva,Francisco Gallegos-Funes.3Dfiltering of colour video sequences using fuzzy logic and vector order statistics[J]Lecture Notes in Computer Science,2009,5807(10):210-221.
[10]龙红梅,田逢春,谭洪涛,任晓羽.一种基于运动估计的3D视频降噪算法[J].计算机与数字工程,2009(6):122-124.
[11] Yau Wai,Lucas Hui. Spatio-temporal video noise reduction system[P].美国,US7145607B1,2006-12-05.
[12] Peng Lin,Yeong-Taeg Kim. Motion adaptive noise reduction apparatus andmethod for video signals[P].美国,US7199838B2,2007-04-03.
[13] Wei-Kuo Lee,Shih-Chung Wang,Yi-Fan Chen. Apparatus and method for temporal noise reduction and motion enhancement[P].美国,US7782402B2,2007-09-24.
[14] Chao-Ho Chen,Ming-Kun Wu. Video noise reduction method using adaptive spatial and motion-compensation temporal filters[P].美国,US7729555B2,2010-06-01.
[15] Boyce J.M. Noise reduction of image sequences using adaptive motion compensated frame averaging[A].IEEE International Conference. Acoustics,Speech,Signal Processing[C].USA,1992,3:461-464.
[16] Lowell L.Winger,Pavel Novotny,Yunwei Jia. Noise adaptive3D composite noisereduction[P].美国,US7551232B2,2009-06-23.
[17] Antoni Buades,Bartomeu Coll,Jean-Michel Morel. Nonlocal image and moviedenoising[J].International Journal of Computer Vision,2008,76(10):123-139.
[18] Ce Liu,William. A high-quality video denoising algorithm based on reliablemotion estimation[J].Lecture Notes in Computer Science,2010,6316(10):706-719.
[19]林书.静态图像、视频以及影片基于内容的高斯降噪[P].中国,CN101213574,2008-07-02.
[20]许海峰,赵文庆,孙峥,陈更生.一种基于降采样的块匹配三步搜索改进算法[J]复旦学报(自然科学版),2010,49(6):653-659.
[21] Song B.C. Motion-compensated temporal filtering for denoising in video encoder[J].Electronics Letters,2004,40(13):802-804.
[22] Jovanov L,Pizurica A,Schulte S,Schelkens P,Munteanu A,Kerre E,Philips W.Combined wavelet-domain and motion-compensated video denoising based onvideo codec motion estimation methods[J].IEEE Transactions on Circuits and Systems for Video Technology,2009,19(3):417-421.
[23]胡少伟,陈景养.用于数字视频信号的运动自适应降噪预处理[P].中国,CN1541482,2004-10-27.
[24]冯鹏,魏彪,米德伶,潘英俊,于渝.基于时域递归滤波的动态数字图像降噪[J].重庆大学学报(自然科学版),2005,28(2):23-25.
[25]黄晓东.一种视频图像3D降噪方法[P].中国,CN1997104,2007-07-11.
[26] Sheng Zhong,San Jose. Method and system for video noise reduction based onmoving content detection[P].美国,US7573530B2,2009-08-11.
[27]黄喆.一种视频降噪的方法和装置[P].中国,CN101001317,2007-07-18.
[28]龙红梅.视频图像降噪算法研究[D].重庆:重庆大学,2009.
[29]张文洁.一种自适应的帧间滤波降噪方法[J].现代电子技术,2007,30(22):130-131.
[30]高辉,谢永昌,狄红卫.一种基于时空联合的实时视频降噪算法[J].微型机与应用,2011,30(16):36-38.
[31] Wade K.Wan,Jason C.Demas,Darren Neuman. Method and system for noisereduction in digital video[P].美国,US7724307B2,2010-05-25.
[32] Kokara A.C. Motion Picture Restoration:Digital Algorithms for Artifact Suppression in Degraded Motion Picture Film and Video[M].英国:Springer-Verla,1998.
[33]冈萨雷斯.数字图像处理[M].北京:电子工业出版社,2007.
[34]崔征.数字视频图像质量评价方法的分析与研究[J].有线电视技术,2008,218(2):93-101.
[35] Olsen S.I. Estimation of noise in images: An evaluation[J].Graphical Models andImage Processing,1993,55(4):319-323.
[36] David L.Donoho,Jain M.Johnstone. Ideal spatial adaptation via wavelet shrinkage[J].Biometrika,1994,81(3):425-455.
[37] Amer A,Dubois E. Fast and reliable structure-oriented video-noise estimation[J].IEEE Transactions on Circuits and System for Video Technology,2005,15(1):113-118.
[38]盛晔,陈东,刘俊义,王丽娟.新的大倾斜摄影的自动快速镶嵌模型[J].计算机工程与设计,2005,26(12):3347-3349.
[39]赖施成,贾泂.基于块内邻域相关度的图像估计噪声[J].计算机与现代化，2009,172(12):82-84.
[40] Buades A,Coll B,Morel J.M. A non-local algorithm for image denoising
[A].IEEE Computer Society Conference. Computer Vision and Pattern Recognition[C].USA:San Diego,2005,2:60-65.
[41] Efstratiadis S.N,Katsaggelos A.K. A model-based pel-recursive motion estimation algorithm[A].International Conference. Acoustics,Speech,and Signal Processing[C].1990,4:1973-1976.
[42] Csillag P,Boroczky L. New methods to improve the accuracy of the pel-recursive wiener-based motion estimation algorithm[A].International Conference.Image Processing[C].1999,3:714-716.
[43] Xiaolin Chen,Canagarajah N,Nunez-Yanez J.H. Backward adaptive pixel-based fast predictive motion estimation[J].IEEE Signal Processing Letters,2009,16(5):370-373.
[44] Jinzenji K,Okada S,Watanabe H,Kobayashi N. Automatic two-layer video objectplane generation scheme and its application to MPEG-4video coding[A].The2000IEEE International Symposium. Circuits and Systems[C].CH:Geneva,2000,3:606-609.
[45] Shyi-Chyi Cheng,Wei-Kan Huang,Tian-Luu Wu. A fast global motion estimation for moving objects segmentation using moment-preserving technique[A].IEEE International Symposium. Circuits and Systems[C].2005,4:3455-3458.
[46]夏金祥,黄顺吉.基于VOP的块特性的自适应十字搜索模式运动估计法[J].通信学报,2005,26(8):117-121.
[47]李志欣.视频编码中的运动估计研究[D].大连:大连理工大学,2006.
[48] Hongiun Jia,Li Zhang. A new cross diamond search algorithm for blockmotion estimation[A].IEEE International Conference. Acoustics,Speech and Signal Processing[C].2004,3:357-360.
[49]魏长虎,贾智平,程志.基于并行和预测的方向菱形运动估计算法[J].计算机应用,2008,28(11):2887-2889.
[50] Ismail Y,McNeelly J,Shaaban M,Bayoumi M.A. Enhanced efficient diamond search algorithm for fast block motion estimation[A].IEEE International Symposium. Circuits and Systems[C].2009:3198-3201.
[51]王镇道,章兢,曾云,陈迪平.基于运动特征的自适应运动估计算法[J].计算机工程与应用,2007,43(31):19-22.
[52]刘显峰.基于结构相似度的图像融合质量评价[D].广州:暨南大学,2007.
[53] Zhou Wang,Alan C Bovik,Ligang Lu. Why is image quality assessment sodifficult[A].2002IEEE International Conference. Acoustics,Speech and Signal Processing[C].2002,4:3313-3316.
[54] Zhou Wang,Ligang Lu,Alan C.Bovik. Video quality assessment based onstrutural distortion measurement[J].Signal Processing: Image Communication,2004,19(2):121-132.
[55] Zhou Wang,Alan C.Bovik. A universal image quality index[J].IEEE SignalProcessing Letters,2002,9(3):81-84.
[56] Zhou Wang,Alan C.Bovik,Ligang Lu. Wavelet-based foveated image quality measurement for region of interest image coding[A].2001International Conference. Image Processing[C].2001,2:89-92.
[2] JingYu Yang. Image and video denoising using adaptive dual-tree discretewavelet packets[J].IEEE Transactions on Circuits and Systems for Video Technology,2009,19(5):642-655.
[3] Luisier F. SURE-LET for orthonormal wavelet-domain video denoising[J].IEEE Transactions on Circuits and Systems for Video Technology,2010,20(6):913-919.
[4] Shigong Yu. Video denoising using motion compensated3-D wavelet transformwith integrated recursive temporal filtering[J].IEEE Transactions on Circuits andSystems for Video Technology,2010,20(6):780-791.
[5]孙延奎.小波分析及其应用[M].北京:机械工业出版社,2005.
[6]程正兴,杨守志,冯晓霞.小波分析的理论、算法、进展和应用[M].北京:国防工业出版社,2007.
[7]安世霞.基于3D小波的视频降噪算法研究[D].兰州:兰州大学,2009.
[8]谭洪涛.视频图像降噪关键技术研究[D].重庆:重庆大学,2010.
[9] Volodymyr Ponomaryov,Alberto Rosales-Silva,Francisco Gallegos-Funes.3Dfiltering of colour video sequences using fuzzy logic and vector order statistics[J]Lecture Notes in Computer Science,2009,5807(10):210-221.
[10]龙红梅,田逢春,谭洪涛,任晓羽.一种基于运动估计的3D视频降噪算法[J].计算机与数字工程,2009(6):122-124.
[11] Yau Wai,Lucas Hui. Spatio-temporal video noise reduction system[P].美国,US7145607B1,2006-12-05.
[12] Peng Lin,Yeong-Taeg Kim. Motion adaptive noise reduction apparatus andmethod for video signals[P].美国,US7199838B2,2007-04-03.
[13] Wei-Kuo Lee,Shih-Chung Wang,Yi-Fan Chen. Apparatus and method for temporal noise reduction and motion enhancement[P].美国,US7782402B2,2007-09-24.
[14] Chao-Ho Chen,Ming-Kun Wu. Video noise reduction method using adaptive spatial and motion-compensation temporal filters[P].美国,US7729555B2,2010-06-01.
[15] Boyce J.M. Noise reduction of image sequences using adaptive motion compensated frame averaging[A].IEEE International Conference. Acoustics,Speech,Signal Processing[C].USA,1992,3:461-464.
[16] Lowell L.Winger,Pavel Novotny,Yunwei Jia. Noise adaptive3D composite noisereduction[P].美国,US7551232B2,2009-06-23.
[17] Antoni Buades,Bartomeu Coll,Jean-Michel Morel. Nonlocal image and moviedenoising[J].International Journal of Computer Vision,2008,76(10):123-139.
[18] Ce Liu,William. A high-quality video denoising algorithm based on reliablemotion estimation[J].Lecture Notes in Computer Science,2010,6316(10):706-719.
[19]林书.静态图像、视频以及影片基于内容的高斯降噪[P].中国,CN101213574,2008-07-02.
[20]许海峰,赵文庆,孙峥,陈更生.一种基于降采样的块匹配三步搜索改进算法[J]复旦学报(自然科学版),2010,49(6):653-659.
[21] Song B.C. Motion-compensated temporal filtering for denoising in video encoder[J].Electronics Letters,2004,40(13):802-804.
[22] Jovanov L,Pizurica A,Schulte S,Schelkens P,Munteanu A,Kerre E,Philips W.Combined wavelet-domain and motion-compensated video denoising based onvideo codec motion estimation methods[J].IEEE Transactions on Circuits and Systems for Video Technology,2009,19(3):417-421.
[23]胡少伟,陈景养.用于数字视频信号的运动自适应降噪预处理[P].中国,CN1541482,2004-10-27.
[24]冯鹏,魏彪,米德伶,潘英俊,于渝.基于时域递归滤波的动态数字图像降噪[J].重庆大学学报(自然科学版),2005,28(2):23-25.
[25]黄晓东.一种视频图像3D降噪方法[P].中国,CN1997104,2007-07-11.
[26] Sheng Zhong,San Jose. Method and system for video noise reduction based onmoving content detection[P].美国,US7573530B2,2009-08-11.
[27]黄喆.一种视频降噪的方法和装置[P].中国,CN101001317,2007-07-18.
[28]龙红梅.视频图像降噪算法研究[D].重庆:重庆大学,2009.
[29]张文洁.一种自适应的帧间滤波降噪方法[J].现代电子技术,2007,30(22):130-131.
[30]高辉,谢永昌,狄红卫.一种基于时空联合的实时视频降噪算法[J].微型机与应用,2011,30(16):36-38.
[31] Wade K.Wan,Jason C.Demas,Darren Neuman. Method and system for noisereduction in digital video[P].美国,US7724307B2,2010-05-25.
[32] Kokara A.C. Motion Picture Restoration:Digital Algorithms for Artifact Suppression in Degraded Motion Picture Film and Video[M].英国:Springer-Verla,1998.
[33]冈萨雷斯.数字图像处理[M].北京:电子工业出版社,2007.
[34]崔征.数字视频图像质量评价方法的分析与研究[J].有线电视技术,2008,218(2):93-101.
[35] Olsen S.I. Estimation of noise in images: An evaluation[J].Graphical Models andImage Processing,1993,55(4):319-323.
[36] David L.Donoho,Jain M.Johnstone. Ideal spatial adaptation via wavelet shrinkage[J].Biometrika,1994,81(3):425-455.
[37] Amer A,Dubois E. Fast and reliable structure-oriented video-noise estimation[J].IEEE Transactions on Circuits and System for Video Technology,2005,15(1):113-118.
[38]盛晔,陈东,刘俊义,王丽娟.新的大倾斜摄影的自动快速镶嵌模型[J].计算机工程与设计,2005,26(12):3347-3349.
[39]赖施成,贾泂.基于块内邻域相关度的图像估计噪声[J].计算机与现代化，2009,172(12):82-84.
[40] Buades A,Coll B,Morel J.M. A non-local algorithm for image denoising
[A].IEEE Computer Society Conference. Computer Vision and Pattern Recognition[C].USA:San Diego,2005,2:60-65.
[41] Efstratiadis S.N,Katsaggelos A.K. A model-based pel-recursive motion estimation algorithm[A].International Conference. Acoustics,Speech,and Signal Processing[C].1990,4:1973-1976.
[42] Csillag P,Boroczky L. New methods to improve the accuracy of the pel-recursive wiener-based motion estimation algorithm[A].International Conference.Image Processing[C].1999,3:714-716.
[43] Xiaolin Chen,Canagarajah N,Nunez-Yanez J.H. Backward adaptive pixel-based fast predictive motion estimation[J].IEEE Signal Processing Letters,2009,16(5):370-373.
[44] Jinzenji K,Okada S,Watanabe H,Kobayashi N. Automatic two-layer video objectplane generation scheme and its application to MPEG-4video coding[A].The2000IEEE International Symposium. Circuits and Systems[C].CH:Geneva,2000,3:606-609.
[45] Shyi-Chyi Cheng,Wei-Kan Huang,Tian-Luu Wu. A fast global motion estimation for moving objects segmentation using moment-preserving technique[A].IEEE International Symposium. Circuits and Systems[C].2005,4:3455-3458.
[46]夏金祥,黄顺吉.基于VOP的块特性的自适应十字搜索模式运动估计法[J].通信学报,2005,26(8):117-121.
[47]李志欣.视频编码中的运动估计研究[D].大连:大连理工大学,2006.
[48] Hongiun Jia,Li Zhang. A new cross diamond search algorithm for blockmotion estimation[A].IEEE International Conference. Acoustics,Speech and Signal Processing[C].2004,3:357-360.
[49]魏长虎,贾智平,程志.基于并行和预测的方向菱形运动估计算法[J].计算机应用,2008,28(11):2887-2889.
[50] Ismail Y,McNeelly J,Shaaban M,Bayoumi M.A. Enhanced efficient diamond search algorithm for fast block motion estimation[A].IEEE International Symposium. Circuits and Systems[C].2009:3198-3201.
[51]王镇道,章兢,曾云,陈迪平.基于运动特征的自适应运动估计算法[J].计算机工程与应用,2007,43(31):19-22.
[52]刘显峰.基于结构相似度的图像融合质量评价[D].广州:暨南大学,2007.
[53] Zhou Wang,Alan C Bovik,Ligang Lu. Why is image quality assessment sodifficult[A].2002IEEE International Conference. Acoustics,Speech and Signal Processing[C].2002,4:3313-3316.
[54] Zhou Wang,Ligang Lu,Alan C.Bovik. Video quality assessment based onstrutural distortion measurement[J].Signal Processing: Image Communication,2004,19(2):121-132.
[55] Zhou Wang,Alan C.Bovik. A universal image quality index[J].IEEE SignalProcessing Letters,2002,9(3):81-84.
[56] Zhou Wang,Alan C.Bovik,Ligang Lu. Wavelet-based foveated image quality measurement for region of interest image coding[A].2001International Conference. Image Processing[C].2001,2:89-92.