计算机科学 ›› 2020, Vol. 47 ›› Issue (5): 172-180.doi: 10.11896/jsjkx.190400060
张墨华1,2, 彭建华1
ZHANG Mo-hua1,2, PENG Jian-hua1
摘要: 近年来,使用高斯混合模型作为块先验的贝叶斯方法取得了优秀的图像复原性能,针对这类模型分量固定及主要依赖外部学习的缺点,提出了一种新的基于狄利克雷过程混合模型的图像先验模型。该模型从干净图像数据库中学习外部通用先验,从退化图像中学习内部先验,借助模型中统计量的可累加性自然实现内外部先验融合。通过聚类的新增及归并机制,模型的复杂度随着数据的增大或缩小而自适应地变化,可以学习到可解释及紧凑的模型。为了求解所有隐变量的变分后验分布,提出了一种结合新增及归并机制的批次更新可扩展变分算法,解决了传统坐标上升算法在大数据集下效率较低、容易陷入局部最优解的问题。在图像去噪及填充实验中,相比传统方法,所提模型无论在客观质量评价还是视觉观感上都更有优势,验证了该模型的有效性。
中图分类号:
[1]JAIN P,TYAGI V.LAPB:Locally adaptive patch-based wavelet domain edge-preserving image denoising[J].Information Scie-nces,2015,294:164-181. [2]QIAO T,REN J,WANG Z,et al.Effective denoising and classification of hyperspectral images using curvelet transform and singular spectrum analysis[J].IEEE Transactions on Geoscience and Remote Sensing,2017,55(1):119-133. [3]DABOV K,FOI A,KATKOVNIK V,et al.Image denoising by sparse 3-D transform-domain collaborative filtering[J].IEEE Transactions on Image Processing,2007,16(8):2080-2095. [4]GAN L,ZHAO F C,YANG M.Self-adaptive group sparse Representation Method for image inpainting[J].Computer Science,2018,45(8):272-276. [5]GAVASKAR R G,CHAUDHURY K N.Fast Adaptive Bilate-ral Filtering[J].IEEE Transactions on Image Processing,2019,28(2):779-790. [6]GUPTA P,MOORTHY A K,SOUNDARARAJAN R,et al.Generalized Gaussian scale mixtures:A model for wavelet coefficients of natural images[J].Signal Processing:Image Communication,2018,66:87-94. [7]XIAO J S,GAO W,PENG H,et al.DetailEnhancement for Ima-ge Super-Resolution Algorithm Based on SVD and Self-Similari-ty[J].Chinese Journal of Computers,2016,39(7):1393-1406. [8]LEBRUN M,BUADES A,MOREL J M.A nonlocal Bayesian image denoising algorithm[J].SIAM Journal on Imaging Scie-nces,2013,6(3):1665-1688. [9]XU J,ZHANG L,ZUO W,et al.Patch group based nonlocalself-similarity prior learning for image denoising[C]//2015 IEEE International Conference on Computer Vision (ICCV).Los Alamitos,CA,USA:IEEE Computer Society,2015:244-252. [10]CHEN F,ZHANG L,YU H.External patch prior guided internal clustering for image denoising[C]//2015 IEEE International Conference on Computer Vision.Los Alamitos,CA,USA:IEEE Computer Society,2015:603-611. [11]ZORAN D,WEISS Y.From learning models of natural image patches to whole image restoration[C]//2011 IEEE International Conference on Computer Vision.Piscataway,NJ,USA:IEEE,2011:479-486. [12]XU J,ZHANG L,ZHANG D.External prior guided internalprior learning for real-world noisy image denoising[J].IEEE Transactions on Image Processing,2018,27(6):2996-3010. [13]ULYANOV D,VEDALDI A,LEMPITSKY V.Deep imageprior[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Los Alamitos,CA,USA:IEEE Computer Society,2018:9446-9454. [14]CHAUDHURY S,ROY H.Can fully convolutional networksperform well for general image restoration problems?[C]//2017 Fifteenth IAPR International Conference on Machine Vision Applications (MVA).Nagoya,Japan:IEEE,2017:254-257. [15]SCHMIDT U,ROTH S.Shrinkage fields for effective image restoration[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition.Los Alamitos,CA,USA:IEEE,2014:2774-2781. [16]CHEN Y,YU W,POCK T.On learning optimized reaction diffusion processes for effective image restoration[C]//2015 IEEE International Conference on Computer Vision.Los Alamitos,CA,USA:IEEE Computer Society,2015:5261-5269. [17]BURGER H C,SCHULER C,HARMELING S.Learning how to combine internal and external denoising methods[C]//German Conference on Pattern Recognition.Berlin:Springer,2013:121-130. [18]MOSSERI I,ZONTAK M,IRANI M.Combining the power of internal and external denoising[C]//IEEE international conference on computational photography (ICCP).Los Alamitos,CA,USA:IEEE Computer Society,2013:1-9. [19]MULLER P,QUINTANA F A,JARA A,et al.Bayesian nonparametric data analysis[M].New York:Springer,2015. [20]PRABHAKARAN S,AZIZI E,CARR A,et al.Dirichlet process mixture model for correcting technical variation in single-cell gene expression data[C]//International Conference on Machine Learning.New York,USA:International Machine Learning Society,2016:1070-1079. [21]ZHANG X.A very gentle note on the construction of dirichlet process[R].Canberra:The Australian National University,2008. [22]SETHURAMAN J.A Constructive Definition of the Dirichlet Prior[J].Statistica Sinica,1994,4(2):639-650. [23]HOSINO T.Two Alternative Criteria for a Split-Merge MCMC on Dirichlet Process Mixture Models[C]//InternationalConfe-rence on Artificial Neural Networks.Cham,Switzerland:Sprin-ger,2017:672-679. [24]BLEI D M,KUCUKELBIR A,MCAULIFFE J D.Variational inference:A review for statisticians[J].Journal of the American Statistical Association,2017,112(518):859-877. [25]HUGHES M C,SUDDERTH E B.Memoized online variational inference for Dirichlet process mixture models[J].Advances in Neural Information Processing Systems,2013:1133-1141. [26]HUGHES M,KIM D I,SUDDERTH E B.Reliable and scalable variational inference for the hierarchical Dirichlet process[C]//18th International Conference on Artificial Intelligence and Statistics.San Diego,CA,USA:Microtome Publishing,2015:370-378. [27]MARTIN D,FOWLKES C,TAL D,et al.A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics[C]//Eighth IEEE International Conference on Computer Vision.Piscataway,NJ,USA:IEEE,2002:416-423. [28]WANG Y Q.E-PLE:An algorithm for image inpainting[J].Ima-ge Processing On Line,2013,3:271-285. |
[1] | 王同森, 史勤忠, 王得法, 董硕, 杨国为, 于腾. 基于光源区域自适应的夜间去雾方法 Nighttime Image Dehazing Method Based on Adaptive Light Source Region 计算机科学, 2021, 48(11A): 327-333. https://doi.org/10.11896/jsjkx.210300072 |
[2] | 李泽文, 李子铭, 费天禄, 王瑞琳, 谢在鹏. 基于残差生成对抗网络的人脸图像复原 Face Image Restoration Based on Residual Generative Adversarial Network 计算机科学, 2020, 47(6A): 230-236. https://doi.org/10.11896/JsJkx.190400118 |
[3] | 邹鹏, 谌雨章, 陈龙彪, 曾张帆. 基于神经网络的光照分布预测夜视复原算法 Night Vision Restoration Algorithm Based on Neural Network for Illumination Distribution Prediction 计算机科学, 2019, 46(11A): 329-333. |
[4] | 张茗琪, 曹国, 陈强, 孙权森. 基于改进逆滤波的衍射成像光谱仪图像复原方法 Image Restoration Method Based on Improved Inverse Filtering for Diffractive Optic Imaging Spectrometer 计算机科学, 2019, 46(1): 86-93. https://doi.org/10.11896/j.issn.1002-137X.2019.01.013 |
[5] | 刘洋, 张杰, 张慧. 一种改进的Retinex算法在图像去雾中的研究与应用 Study and Application of Improved Retinex Algorithm in Image Defogging 计算机科学, 2018, 45(6A): 242-243. |
[6] | 许影,李强懿. 基于稀疏特性的盲二值图像去模糊 Blind Binary Image Deconvolution Based on Sparse Property 计算机科学, 2018, 45(3): 253-257. https://doi.org/10.11896/j.issn.1002-137X.2018.03.040 |
[7] | 刘付勇,高贤强,张著. 基于改进贝叶斯概率模型的推荐算法 Improved Bayesian Probabilistic Model Based Recommender System 计算机科学, 2017, 44(5): 285-289. https://doi.org/10.11896/j.issn.1002-137X.2017.05.052 |
[8] | 马洪华,黄永林,丁岩岩. 用于彩色图像复原的带有高阶耦合项的TV模型 Total Variance with High-order Coupling Term for Color Image Restoration 计算机科学, 2016, 43(Z6): 214-216. https://doi.org/10.11896/j.issn.1002-137X.2016.6A.051 |
[9] | 赵志刚,陈莹莹,赵 毅,张维忠,吕慧显,潘振宽. 基于边缘先验模型的运动去模糊 Motion Deblurring Based on Edge Prior Model 计算机科学, 2015, 42(5): 305-308. https://doi.org/10.11896/j.issn.1002-137X.2015.05.062 |
[10] | 刘建伟,崔立鹏,黎海恩,罗雄麟. 概率图模型推理方法的研究进展 Research and Development on Inference Technique in Probabilistic Graphical Models 计算机科学, 2015, 42(4): 1-18. https://doi.org/10.11896/j.issn.1002-137X.2015.04.001 |
[11] | 肖泉,丁兴号,廖英豪. 一种有效保持边缘特征的散焦模糊图像复原方法 Novel Edge-preserving Algorithm for Defocus Blurred Image Restoration 降 计算机科学, 2010, 37(7): 270-272. |
[12] | 肖宿,韩国强,沃焱. 多信道图像盲复原算法 Multichannel Blind Image Restoration 计算机科学, 2010, 37(12): 234-237. |
|