在线阅读 --自然科学版 2019年2期《一种基于扩散驱动先验技术的图像超分辨率重构方法》
一种基于扩散驱动先验技术的图像超分辨率重构方法--[在线阅读]
陈孟臻1, 陈莹1, 卢振坤2
1. 百色学院 信息工程学院, 广西 百色 533000;
2. 广西民族大学 信息科学与工程学院, 广西 南宁 530006
起止页码: 104--111页
DOI: 10.13763/j.cnki.jhebnu.nse.2019.02.003
摘要
当输入图像因污迹、噪声和采样而严重退化时,目前基于Papoulis-Gerchberg(PG)算法的大多数超分辨率方法表现不佳.因此,提出了一种基于扩散驱动先验和PG算法的超分辨率方法,能够在提高图像分辨率的同时,估计缺失的高频分量.首先提出了一种新型扩散驱动平滑的先验,能够在平坦和轮廓区域之间自动平衡作用,确保正则化水平以产生清晰图像.然后,将PG算法引入到迭代过程中,以估计重构场景中缺失的小规模特征.实验结果表明,相比现有的超分辨率方法,提出方法的峰值信噪比和结构相似指数结果更高,重构图像更加清晰且无伪影.

An Image Super-resolution Reconstruction Method Based on Diffusion Driven Prior
CHEN Mengzhen1, CHEN Ying1, LU Zhenkun2
1. School of Information Engineering, Baise University, Guangxi Baise 533000, China;
2. School of Information Science and Engineering, Guangxi University for Nationalities, Guangxi Nanning 530006, China
Abstract:
Most of the super-resolution methods currently based on the Papoulis-Gerchberg (PG) algorithm perform poorly when input images are heavily degraded by smudge,noise and sampling.Therefore, a super-resolution method based on diffusion driven priors and PG algorithms is proposed to estimate the missing high-frequency components while improving image resolution.First,a novel diffusion driven smoothing priors that can automatically balance between flat and contour regions are developed to ensure regularization levels to produce clearer images.Then,the PG algorithm is introduced into the iterative process to estimate the small scale features in the reconstructed scene.The experimental results show that compared with the existing super-resolution methods,the proposed method has higher peak signal-to-noise ratio(PSNR) and structure similarity index(SSIM),and the reconstructed image is clearer and without artifacts.

收稿日期: 2018-10-15
基金项目: 国家自然科学基金(61561008);广西壮族自治区教育厅高校中青年教师基础能力提升项目(KY2016LX339);百色学院应用型本科专业重点建设项目(2015YYZY01)

参考文献:
[1]YUE L,SHEN H,YUAN Q,et al.A Locally Adaptive L1-L2 Norm for Multi-frame Super-resolution of Images with Mixed Noise and Outliers[J].Signal Processing,2014,105(12):156-174.doi:10.1016/j.sigpro.2014.04.031
[2]DENG L J,GUO W,HUANG T Z.Single-image Super-resolution via an Iterative Reproducing Kernel Hilbert Space Method[J].IEEE Transactions on Circuits & Systems for Video Technology,2016,26(11):2001-2014.doi:10.1109/TCSVT.2015.2475895
[3]莫建文,曾儿孟,张彤,等.基于多字典学习和图像块映射的超分辨率重建[J].计算机应用,2016,36(5):1394-1398.doi:10.11772/j.issn.1001-9081.2016.05.1394 MO Jianwen,ZENG Ermeng,ZHANG Tong,et al.Super-resolution Reconstruction Based on Multi Dictionary Learning and Image Block Mapping[J].Computer Applications,2016,36(5):1394-1398.
[4]MAISELI B J,ELISHA O A,GAO H.A Multi-frame Super-resolution Method Based on the Variable-exponent Nonlinear Diffusion Regularizer[J].Eurasip Journal on Image & Video Processing,2015,2015(1):22.doi:10.1186/s13640-015-007-2
[5]REN Z,HE C,ZHANG Q.Fractional Order Total Variation Regularization for Image Super-resolution[J].Signal Processing,2013,93(9):2408-2421.doi:10.1016/j.sigpro.2013.02.015
[6]LI X,HU Y,GAO X,et al.A Multi-frame Image Super-resolution Method[J].Signal Processing,2010,90(2):405-414.doi:10.1016/j.sigpro.2009.05.028
[7]YUAN Q,ZHANG L,SHEN H.Multiframe Super-resolution Employing a Spatially Weighted Total Variation Model[J].IEEE Transactions on Circuits & Systems for Video Technology,2012,22(3):379-392.doi:10.1109/tcsvt.2011.2163447
[8]ZENG W,LU X,FEI S.Image Super-resolution Employing a Spatial Adaptive Prior Model[J].Neurocomputing,2015,162(C):218-233.doi:10.1016/j.neucom.2015.03.049
[9]CHOUZENOUX E,JEZIERSKA A,PESQUET J C,et al.A Majorize-minimize Subspace Approach for L2-L0 Image Regularization[J].Siam Journal on Imaging Sciences,2013,6(1):563-591.doi:10.1137/11085997X
[10]LOPEZ-MOLINA C,DE BAETS B,CERRON J,et al.A Generalization of the Perona-malik Anisotropic Diffusion Method Using Restricted Dissimilarity Functions[J].International Journal of Computational Intelligence Systems,2013,6(1):14-28.doi:10.1080/18756891.2013.752658
[11]WEICKERT J,ROMENY B H,VIERGEVER M A.Efficient and Reliable Schemes for Nonlinear Diffusion Filtering[J].IEEE Transactions on Image Processing a Publication of the IEEE Signal Processing Society,1998,7(3):398-410.doi:10.1109/83.661190
[12]PERONA P,MALIK J.Scale-space and Edge Detection Using Anisotropic Diffusion[J].IEEE Trans Pattern Anal Mach Intell,1990,12(7):629-639.doi:10.1109/34.56205
[13]WEICKERT J,ROMENY B H,VIERGEVER M A.Efficient and Reliable Schemes for Nonlinear Diffusion Filtering[J].IEEE Transactions on Image Processing,1998,7(3):398-410.doi:10.1109/83.661190
[14]FARSIU S,ROBINSON D,ELAD M,et al.Advances and Challenges in Super[J].International Journal of Imaging Systems & Technology,2004,14(2):47-57.doi:10.1002/ima.20007
[15]岳妍.基于可信特征距离变换的模型匹配算法研究[D].武汉:华中科技大学,2015.YUE Yan.Research on Model Matching Algorithm Based on Distance Transformation of Trusted Feature[D].Wuhan:Huazhong University of Science and Technology,2015.
[16]SALARI E,BAO G.Super-resolution Using an Enhanced Papoulis-gerchberg Algorithm[J].Image Processing Iet,2012,6(7):959-965.doi:10.1049/iet-ipr.2010.0513
[17]GHOLIPOUR A,ESTROFF J A,WARFIELD S K.Robust Super-resolution Volume Reconstruction from Slice Acquisitions:Application to Fetal Brain MRI[J].IEEE Transactions on Medical Imaging,2010,29(10):1739-1758.doi:10.1109/TMI.2010.2051680
[18]姜蕴芝,蔡志权,葛永斌.求解一维抛物型方程的一类半隐格式及组显格式[J].兰州文理学院学报(自然科学版),2014,28(3):7-10.doi:10.3969/j.issn.2095-6991.2014.03.003 JIANG Yunzhi,CAI Zhiquan,GE Yongbin.A Class of Semi Implicit Schemes for Solving One-dimensional Parabolic Equations and Group Explicit Schemes[J].Journal of Lanzhou College of Arts and Sciences(Naturl Science Edition),2014,28(3):7-10.
[19]LAGHRIB A,HAKIM A,RAGHAY S.A Combined Total Variation and Bilateral Filter Approach for Image Robust Super Resolution[J].Eurasip Journal on Image & Video Processing,2015,2015(1):19-26.doi:10.1186/s13640-015-0075-4