基函数神经网络图像复原算法研究
|
摘要
图像复原技术是图像处理领域重要的研究课题之一。在天文遥感军事、道路交通及侦破、医学影像等领域具有重要的现实意义和应用价值。图像的复原技术属于图像研究的预处理部分，也是最基础的环节，复原结果的好坏对数字图像的后续研究工作有着重要的地位。然而在现实中，获取图像时，会由于各种因素造成图像的退化。因此如何复原出退化的数字图像是近年来数字图像处理领域研究的热点问题。
由于现阶段造成图像退化的因素复杂多样，图像的退化函数难以准确地确定，论文中提出了一种利用基函数神经网络的图像复原算法，它是以一组正交多项式函数作为隐层神经元的激励函数，根据误差传递算法（BP算法）对权值进行修正，达到收敛目标。由于神经网络的拓扑结构对网络的性能影响极大，文中利用了衍生算法确定神经网络在给定误差下得出处于最佳性能时的隐层神经元的数目，针对传统的神经网络学习需要进行反复的迭代，文中寻找出在基函数神经网络的迭代次数与网络权值的关系，提出了权值直接确定的图像复原算法，相对于传统算法，具有较好的效果。
针对局部退化图像的问题，论文中提出了一种利用图像的频谱特征进行图像退化区域提取的方法，与传统方法相比，该方法不受背景区域的影响，能较好的完成图像完全退化区域的提取。
Image restoration is one of the important research topics in the field ofimage processing. It has important practical significance and application valuein military remote sensing, Road traffic, detection and medical imaging and soon. The technology of image restoration technology belongs to imagepreprocessing, but the fundamental part. The results of the restoration of digitalimages play an important role in the further research. However, it would bedegraded when we obtain the image in reality. So how to recover the degradedimages is becoming a hot issue in recent years.
Image degradation is caused by complex and diverse factors, and the pointspread function is difficult to determine. According to the fact that point spreadfunction of the degraded image can’t obtain accurately, basis function neuralnetwork for image restoration was constructed based on the Orthogonalpolynomial basis functions in this paper, The hidden-layer neurons are activatedby a series of orthogonal functions, update its weights by the errorBack-propagation training algorithm and finally reached convergence target. Asthe topology of the neural network take a great impact on its performance, thispaper determine the number of the hidden layer neurons which under the bestperformance in a given error using hidden-neuron growing algorithm. To avoidlengthy BP-training of the iterative weights-up dating, this paper found therelationship between the number of iterative training and the weights,weights-direct-determination algorithm for image restoration was proposed,Compared with traditional algorithms, it is has good results.
Finally, A new method for region extraction of partial blurred image ispresented based on the Fourier spectrum amplitudes, this method can avoid theaffect of background area, realize the extract blur region well compare to thetraditional methods.
由于现阶段造成图像退化的因素复杂多样，图像的退化函数难以准确地确定，论文中提出了一种利用基函数神经网络的图像复原算法，它是以一组正交多项式函数作为隐层神经元的激励函数，根据误差传递算法（BP算法）对权值进行修正，达到收敛目标。由于神经网络的拓扑结构对网络的性能影响极大，文中利用了衍生算法确定神经网络在给定误差下得出处于最佳性能时的隐层神经元的数目，针对传统的神经网络学习需要进行反复的迭代，文中寻找出在基函数神经网络的迭代次数与网络权值的关系，提出了权值直接确定的图像复原算法，相对于传统算法，具有较好的效果。
针对局部退化图像的问题，论文中提出了一种利用图像的频谱特征进行图像退化区域提取的方法，与传统方法相比，该方法不受背景区域的影响，能较好的完成图像完全退化区域的提取。
Image restoration is one of the important research topics in the field ofimage processing. It has important practical significance and application valuein military remote sensing, Road traffic, detection and medical imaging and soon. The technology of image restoration technology belongs to imagepreprocessing, but the fundamental part. The results of the restoration of digitalimages play an important role in the further research. However, it would bedegraded when we obtain the image in reality. So how to recover the degradedimages is becoming a hot issue in recent years.
Image degradation is caused by complex and diverse factors, and the pointspread function is difficult to determine. According to the fact that point spreadfunction of the degraded image can’t obtain accurately, basis function neuralnetwork for image restoration was constructed based on the Orthogonalpolynomial basis functions in this paper, The hidden-layer neurons are activatedby a series of orthogonal functions, update its weights by the errorBack-propagation training algorithm and finally reached convergence target. Asthe topology of the neural network take a great impact on its performance, thispaper determine the number of the hidden layer neurons which under the bestperformance in a given error using hidden-neuron growing algorithm. To avoidlengthy BP-training of the iterative weights-up dating, this paper found therelationship between the number of iterative training and the weights,weights-direct-determination algorithm for image restoration was proposed,Compared with traditional algorithms, it is has good results.
Finally, A new method for region extraction of partial blurred image ispresented based on the Fourier spectrum amplitudes, this method can avoid theaffect of background area, realize the extract blur region well compare to thetraditional methods.
引文
[1]张秉仁,陈里铭,高游.运动模糊图像的降质过程分析与恢复技术研究[J].中国图象图形学报,2004,9(7):815-819.
[2]朱俊.图像复原技术及其在数字相机成像品质改善中的应用[D].重庆大学,2004.
[3] K.R Castleman.Digital Image Processing. Prentice Hall.1998,9:255-278.
[4] HARRIS, S.R.J.L., Image evaluation and restoration. JOSA,1966.56(5):569-570.
[5] Helstrom, C.W., Image restoration by the method of least squares. JOSA,1967.57(3):297-303.
[6] T.M Cannon. Digital Image Deblurring by Nonliner Homomorphic Filtering[D].Ph.D.Thesis,Computer science Department, university of Utah,Salt Lake City,1974.
[7] B.R Hunt. The application of constrained least squares estimation to image restoration bydigital computer[C].IEEE Trans,1973. C-22(9):805-812.
[8]Donoho, D.L. and J.M. Johnstone. Ideal spatial adaptation by wavelet shrinkage[J].Biometrika,1994.81(3):425-455.
[9] Shark, L.K. and C. Yu, Denoising by optimal fuzzy thresholding in wavelet domain.Electronics Letters,2000.36(6):581-582.
[10]Dempster, A.P., N.M. Laird, and D.B. Rubin, Maximum likelihood from incomplete datavia the EM algorithm. Journal of the Royal Statistical Society. Series B (Methodological),1977:1-38.
[11]Figueiredo, M.A.T. R.D. Nowak. An EM algorithm for wavelet-based imagerestoration[J]. IEEE Transactions on Image Processing,2003.12(8):906-916.
[12]Galatsanos, N.P., et al., Hierarchical Bayesian image restoration from partially knownblurs. Image Processing, IEEE Transactions on,2000.9(10):1784-1797.
[13]闫河,闫卫军,李唯唯.基于Lucy-Richardson算法的图像复原[J].计算机工程,2010,36(015):204-210.
[14]沈瑛,吴禄慎.由约束最小二乘方法改进的图像恢复方法[J].数据采集与处理,2002.17(3):325-327.
[15]陈春涛,黄步根.残损模糊图像的最大熵恢复[J].计算机应用与软件,2004，21(8):19-20.
[16]汪雪林,韩华,彭思龙.基于小波域局部高斯模型的图像复原[J].软件学报,2004，15(3):443-449.
[17]吴振宇,姚洪利,杜少军.运动模糊参数的空域鉴别方法[J].计算机应用,2009，29(12):3378-3380.
[18]许元男,赵远,刘丽萍等.基于双谱的点扩散函数参数辨识[J].光电工程，2009,36(5):72-76.
[19]邹谋炎.反卷积和信号复原[M].国防工业出版社,2004.
[20]Kingsbury, N., Complex wavelets for shift invariant analysis and filtering of signals[J].Applied and Computational Harmonic Analysis,2001,10(3):234-253.
[21]Zhou, Y.T., et al., Image restoration using a neural network[J]. IEEE Transactions onAcoustics, Speech and Signal Processing,1988.36(7):1141-1151.
[22]Paik, J.K. and A.K. Katsaggelos. Image restoration using a modified Hopfield network[J].IEEE Transactions on Image Processing,1992.1(1):49-63.
[23]Cha, I. and S.A. Kassam. RBFN restoration of nonlinearly degraded images[J]. IEEETransactions on Image Processing,1996,5(6):964-975.
[24]Celebi M E,Giizelis C.image restoration using cellular neural network[J].ElectronicsLeters,1997,33(1):43-45.
[25]Clarke L P, Qian W. Fuzzy-logic adaptive neural network for nuclear nedicine imagerestoration[C].the20th Annual Internation Conference on Engineering in Medicine andBiology Society,1998,3:1363-1366.
[26]汪海明,赵建业,程承旗等.细胞神经网络图像恢复新方法的研究[J].计算机工程与应用,2003,39(6):4-6
[27]刘普寅,李洪兴.基于模糊神经网络的图像恢复技术[J].中国科学(E),2002,32(4)：541-552.
[28]Qian, W. and L.P. Clarke, Wavelet-based Neural Network with Fuzzy-logic Adaptivityfor Nuclear Image Restoration. Proceedings of the IEEE,1996,84(10):1458-1473.
[29]朱策,杨小帆,陈静.一种新的图像恢复遗传算法[J].计算机应用,2006,26(6):1368-1369.
[30]刘志军,丁明跃,周成平.基于并行遗传算法的图像超分辨率复原[J].中国图象图形学报,2004,9(1):62-68.
[31]董俊.基于BP神经网络的图像复原算法研究[D].西安科技大学,2009.
[32]王宇,邹强.基于最小二乘支持向量机的模糊图像恢复[J].微型机与应用,2009,28(24):53-55.
[33]贺可鑫,何小海,陶青川.基于RBF神经网络的COSM图像复原算法[J].计算机应用,2009,9(1):78-85
[34]杨宇光,王叶红,王园.基于遗传算法和LM优化的BP神经网络的图像复原算法[J].微计算机应用,2010,31(10):7-13.
[35]Guo, P., H. Li, and M.R. Lyu, Blind image restoration by combining wavelet transformand RBF neural network. International Journal of Wavelets Multiresolution and InformationProcessing,2007.5(1):1-12.
[36]Perry, S.W. and L. Guan, Weight assignment for adaptive image restoration by neuralnetworks[J]. IEEE Transactions on Neural Networks,2000,11(1):156-170.
[37]王菲.运动模糊图像的恢复及恢复质量评价[D].西安电子科技大学,2010.
[38]邹阿金,张雨浓.基函数神经网络及应用[M].广州:中山大学出版社,2009.
[39]吴小俊,王士同,杨静宇.基于正交多项式函数的神经网络及其性质研究[J].计算机工程与应用.2002,38(9):25-26.
[40]曹飞龙,徐宗本,梁吉业.多项式函数的神经网络逼近:网络的构造与逼近算法[J].计算机学报,2003.26(008)：906-912.
[41]张日东,王树青,基于神经网络的非线性系统多步预测控制[J].控制与决策,2005,20(3):332-336.
[42]邹阿金,沈建中.基于Chebyshev神经网络的非线性预测应用研究[J].计算机应用2001,21(4):14-15.
[43]曹飞龙,张永全,潘星.构造前向神经网络逼近多项式函数[J].模式识别与人工智能，2007,20(3):331-335.
[44]丛爽,向微. BP网络结构、参数及训练方法的设计与选择.计算机工程,2001.27(10):36-38.
[45]Mishra, S. K., G. Panda, S. Meher. Chebyshev Functional Link Artificial NeuralNetworks for Denoising of Image Corrupted by Salt and Pepper Noise[J]. InternationalJournal,2009,1(1):413-417.
[46]Purwar, S, I. Kar, A. Jha. On-line system identification of complex systems usingChebyshev neural networks. Applied Soft Computing,2007.7(1):364-372.
[47]肖秀春,姜孝华,张雨浓.一种基函数神经网络最优隐神经元数目快速确定算法[J].微电子学与计算机,2010,27(1):57-60.
[48] Tan, H., Fourier neural networks and generalized single hidden layer networks in aircraftengine fault diagnostics[J]. Journal of Engineering for Gas turbines and Power.2006.128(4):773-782.
[49]张雨浓,曾庆淡,肖秀春等.复指数Fourier神经元网络隐神经元衍生算法[J].计算机应用,2008.28(10):2504-250
[50]张雨浓,旷章辉,肖秀春,陈柏桃. Fourier三角基神经元网络的权值直接确定法[J].计算机工程与科学.2009,31(5):112-115.
[51]Halawa, K., Determining the Weights of A Fourier Series Neural Network on the Basis ofthe Multidimensional Discrete Fourier Transform[J]. International Journal of AppliedMathematics and Computer Science.2008.18(3):369-375.
[52]张雨浓,李巍,蔡炳煌等.切比雪夫正交基神经网络的权值直接确定法[J].计算机仿真,2009,26(1):157-161.
[53]景晓军,李剑锋,刘郁林.一种基于最大类间方差的图像分割算法[J].电子学报,2003,31(9):1281-1285.
[54]袁杰,都思丹,高敦堂.高阶统计量在运动目标检测中的研究[J].模式识别与人工智能,2006,19(1):84-88.
[55]张玉叶,周晓东,王春歆.应用像素运动模糊特征分割的空间移变降质复原[J].光学精密工程,2009,17(5):1119-1125.
[56]Reddick,W.E.,etal.,Automated segmentation and classification multispectral magneticresonance images of brain using artificial Neural Networks[J].IEEE Transactions onmedical imaging,1997,16(6):911-918.
[57]尹兵,王延斌,刘威.利用神经网络鉴别退化图像的模糊类型[J].光学技术,2006.32(1):138-140.
[58]Sarimveis, H., P. Doganis, and A. Alexandridis, A classification technique based onradial basis function neural networks[J]. Advances in Engineering Software,2006,37(4):218-221.
[59]Yi, Z., Foundations of implementing the competitive layer model by Lotka¨CVolterrarecurrent neural networks [J]. IEEE Transactions on Neural Networks,2010,21(3):494-507.
[60]Mao, K., K. C. Tan, et al."Probabilistic neural-network structure determination forpattern classification[J]. IEEE Transactions on Neural Networks.2000,11(4):1009-1016.
[2]朱俊.图像复原技术及其在数字相机成像品质改善中的应用[D].重庆大学,2004.
[3] K.R Castleman.Digital Image Processing. Prentice Hall.1998,9:255-278.
[4] HARRIS, S.R.J.L., Image evaluation and restoration. JOSA,1966.56(5):569-570.
[5] Helstrom, C.W., Image restoration by the method of least squares. JOSA,1967.57(3):297-303.
[6] T.M Cannon. Digital Image Deblurring by Nonliner Homomorphic Filtering[D].Ph.D.Thesis,Computer science Department, university of Utah,Salt Lake City,1974.
[7] B.R Hunt. The application of constrained least squares estimation to image restoration bydigital computer[C].IEEE Trans,1973. C-22(9):805-812.
[8]Donoho, D.L. and J.M. Johnstone. Ideal spatial adaptation by wavelet shrinkage[J].Biometrika,1994.81(3):425-455.
[9] Shark, L.K. and C. Yu, Denoising by optimal fuzzy thresholding in wavelet domain.Electronics Letters,2000.36(6):581-582.
[10]Dempster, A.P., N.M. Laird, and D.B. Rubin, Maximum likelihood from incomplete datavia the EM algorithm. Journal of the Royal Statistical Society. Series B (Methodological),1977:1-38.
[11]Figueiredo, M.A.T. R.D. Nowak. An EM algorithm for wavelet-based imagerestoration[J]. IEEE Transactions on Image Processing,2003.12(8):906-916.
[12]Galatsanos, N.P., et al., Hierarchical Bayesian image restoration from partially knownblurs. Image Processing, IEEE Transactions on,2000.9(10):1784-1797.
[13]闫河,闫卫军,李唯唯.基于Lucy-Richardson算法的图像复原[J].计算机工程,2010,36(015):204-210.
[14]沈瑛,吴禄慎.由约束最小二乘方法改进的图像恢复方法[J].数据采集与处理,2002.17(3):325-327.
[15]陈春涛,黄步根.残损模糊图像的最大熵恢复[J].计算机应用与软件,2004，21(8):19-20.
[16]汪雪林,韩华,彭思龙.基于小波域局部高斯模型的图像复原[J].软件学报,2004，15(3):443-449.
[17]吴振宇,姚洪利,杜少军.运动模糊参数的空域鉴别方法[J].计算机应用,2009，29(12):3378-3380.
[18]许元男,赵远,刘丽萍等.基于双谱的点扩散函数参数辨识[J].光电工程，2009,36(5):72-76.
[19]邹谋炎.反卷积和信号复原[M].国防工业出版社,2004.
[20]Kingsbury, N., Complex wavelets for shift invariant analysis and filtering of signals[J].Applied and Computational Harmonic Analysis,2001,10(3):234-253.
[21]Zhou, Y.T., et al., Image restoration using a neural network[J]. IEEE Transactions onAcoustics, Speech and Signal Processing,1988.36(7):1141-1151.
[22]Paik, J.K. and A.K. Katsaggelos. Image restoration using a modified Hopfield network[J].IEEE Transactions on Image Processing,1992.1(1):49-63.
[23]Cha, I. and S.A. Kassam. RBFN restoration of nonlinearly degraded images[J]. IEEETransactions on Image Processing,1996,5(6):964-975.
[24]Celebi M E,Giizelis C.image restoration using cellular neural network[J].ElectronicsLeters,1997,33(1):43-45.
[25]Clarke L P, Qian W. Fuzzy-logic adaptive neural network for nuclear nedicine imagerestoration[C].the20th Annual Internation Conference on Engineering in Medicine andBiology Society,1998,3:1363-1366.
[26]汪海明,赵建业,程承旗等.细胞神经网络图像恢复新方法的研究[J].计算机工程与应用,2003,39(6):4-6
[27]刘普寅,李洪兴.基于模糊神经网络的图像恢复技术[J].中国科学(E),2002,32(4)：541-552.
[28]Qian, W. and L.P. Clarke, Wavelet-based Neural Network with Fuzzy-logic Adaptivityfor Nuclear Image Restoration. Proceedings of the IEEE,1996,84(10):1458-1473.
[29]朱策,杨小帆,陈静.一种新的图像恢复遗传算法[J].计算机应用,2006,26(6):1368-1369.
[30]刘志军,丁明跃,周成平.基于并行遗传算法的图像超分辨率复原[J].中国图象图形学报,2004,9(1):62-68.
[31]董俊.基于BP神经网络的图像复原算法研究[D].西安科技大学,2009.
[32]王宇,邹强.基于最小二乘支持向量机的模糊图像恢复[J].微型机与应用,2009,28(24):53-55.
[33]贺可鑫,何小海,陶青川.基于RBF神经网络的COSM图像复原算法[J].计算机应用,2009,9(1):78-85
[34]杨宇光,王叶红,王园.基于遗传算法和LM优化的BP神经网络的图像复原算法[J].微计算机应用,2010,31(10):7-13.
[35]Guo, P., H. Li, and M.R. Lyu, Blind image restoration by combining wavelet transformand RBF neural network. International Journal of Wavelets Multiresolution and InformationProcessing,2007.5(1):1-12.
[36]Perry, S.W. and L. Guan, Weight assignment for adaptive image restoration by neuralnetworks[J]. IEEE Transactions on Neural Networks,2000,11(1):156-170.
[37]王菲.运动模糊图像的恢复及恢复质量评价[D].西安电子科技大学,2010.
[38]邹阿金,张雨浓.基函数神经网络及应用[M].广州:中山大学出版社,2009.
[39]吴小俊,王士同,杨静宇.基于正交多项式函数的神经网络及其性质研究[J].计算机工程与应用.2002,38(9):25-26.
[40]曹飞龙,徐宗本,梁吉业.多项式函数的神经网络逼近:网络的构造与逼近算法[J].计算机学报,2003.26(008)：906-912.
[41]张日东,王树青,基于神经网络的非线性系统多步预测控制[J].控制与决策,2005,20(3):332-336.
[42]邹阿金,沈建中.基于Chebyshev神经网络的非线性预测应用研究[J].计算机应用2001,21(4):14-15.
[43]曹飞龙,张永全,潘星.构造前向神经网络逼近多项式函数[J].模式识别与人工智能，2007,20(3):331-335.
[44]丛爽,向微. BP网络结构、参数及训练方法的设计与选择.计算机工程,2001.27(10):36-38.
[45]Mishra, S. K., G. Panda, S. Meher. Chebyshev Functional Link Artificial NeuralNetworks for Denoising of Image Corrupted by Salt and Pepper Noise[J]. InternationalJournal,2009,1(1):413-417.
[46]Purwar, S, I. Kar, A. Jha. On-line system identification of complex systems usingChebyshev neural networks. Applied Soft Computing,2007.7(1):364-372.
[47]肖秀春,姜孝华,张雨浓.一种基函数神经网络最优隐神经元数目快速确定算法[J].微电子学与计算机,2010,27(1):57-60.
[48] Tan, H., Fourier neural networks and generalized single hidden layer networks in aircraftengine fault diagnostics[J]. Journal of Engineering for Gas turbines and Power.2006.128(4):773-782.
[49]张雨浓,曾庆淡,肖秀春等.复指数Fourier神经元网络隐神经元衍生算法[J].计算机应用,2008.28(10):2504-250
[50]张雨浓,旷章辉,肖秀春,陈柏桃. Fourier三角基神经元网络的权值直接确定法[J].计算机工程与科学.2009,31(5):112-115.
[51]Halawa, K., Determining the Weights of A Fourier Series Neural Network on the Basis ofthe Multidimensional Discrete Fourier Transform[J]. International Journal of AppliedMathematics and Computer Science.2008.18(3):369-375.
[52]张雨浓,李巍,蔡炳煌等.切比雪夫正交基神经网络的权值直接确定法[J].计算机仿真,2009,26(1):157-161.
[53]景晓军,李剑锋,刘郁林.一种基于最大类间方差的图像分割算法[J].电子学报,2003,31(9):1281-1285.
[54]袁杰,都思丹,高敦堂.高阶统计量在运动目标检测中的研究[J].模式识别与人工智能,2006,19(1):84-88.
[55]张玉叶,周晓东,王春歆.应用像素运动模糊特征分割的空间移变降质复原[J].光学精密工程,2009,17(5):1119-1125.
[56]Reddick,W.E.,etal.,Automated segmentation and classification multispectral magneticresonance images of brain using artificial Neural Networks[J].IEEE Transactions onmedical imaging,1997,16(6):911-918.
[57]尹兵,王延斌,刘威.利用神经网络鉴别退化图像的模糊类型[J].光学技术,2006.32(1):138-140.
[58]Sarimveis, H., P. Doganis, and A. Alexandridis, A classification technique based onradial basis function neural networks[J]. Advances in Engineering Software,2006,37(4):218-221.
[59]Yi, Z., Foundations of implementing the competitive layer model by Lotka¨CVolterrarecurrent neural networks [J]. IEEE Transactions on Neural Networks,2010,21(3):494-507.
[60]Mao, K., K. C. Tan, et al."Probabilistic neural-network structure determination forpattern classification[J]. IEEE Transactions on Neural Networks.2000,11(4):1009-1016.