当前位置>主页 > 期刊在线 > 计算机技术 >

计算机技术2020年22期

基于模糊核结构约束的图像盲去模糊算法
岳灿¹,官斌²,杨栩萌³
(1. 中国人民武装警察部队湖北省总队,湖北 武汉 430070;2. 海军装备部驻武汉地区第七军事代表室,湖北 武汉 430205; 3. 江苏自动化研究所,江苏 连云港 222061)

摘  要:模糊核估计在图像盲去模糊问题中十分重要。然而,目前大多数方法都没有考虑到利用模糊核本身的结构信息。针对这一问题,文章首先通过观察发现结构较为集中的模糊核可以恢复出更高质量的图像,然后设计了一种使得模糊核结构更集中的约束项,最后基于该约束项提出了一种图像盲去模糊的模型。实验结果表明,该方法相较于其他方法在峰值信噪比性能上有较大提升,可以更好地恢复出图像细节与纹理。


关键词:图像盲去模糊;模糊核估计;结构先验;变分模型



中图分类号:TP391.41         文献标识码:A         文章编号:2096-4706(2020)22-0103-05


Image Blind Deblurring Algorithm Based on Fuzzy Kernel Structure Constraint

YUE Can1,GUAN Bin2,YANG Xumeng3

(1.Hubei Provincial Corps of Chinese People’s Armed Police Force,Wuhan 430070,China;

2.The Seventh Military Representative Office of Naval Equipment Department in Wuhan,Wuhan 430205,China;3.Jiangsu Automation Research Institute,Lianyungang 222061,China)

Abstract:Fuzzy kernel estimation is very important in image blind deblurring. However,most of the current methods do notconsider using the structure information of fuzzy kernel itself. In response to this problem,the article first found through observation thatthe fuzzy kernel with a more concentrated structure can restore a higher quality image,and then designed a constraint that makes the fuzzykernel structure more concentrated,and finally proposed a image blind deblurring model is based on the constraint. Experimental resultsshow that compared with other methods,this method has a greater improvement in peak signal-to-noise ratio performance,and can betterrestore image details and textures.

Keywords:image blind deblurring;fuzzy kernel estimation;structure prior;variational model


参考文献:

[1] WHYTE O,SIVIC J,ZISSERMAN A,et al. Non-uniformDeblurring for Shaken Images [J].International Journal of ComputerVision,2012,98(2):168-186.

[2] LEVIN A,FERGUS R,DURAND F,et al. Image and depthfrom a conventional camera with a coded aperture [J].ACM Transactionson Graphics(TOG),2007,26(3):70.

[3] KRISHNAN D,TAY T,FERGUS R. Blinddeconvolution using a normalized sparsity measure [C]//Conference on Computer Vision and Pattern Recognition(CVPR).IEEE,2011:233-240.

[4] 方帅,刘远东,曹洋,等. 基于模糊结构图的模糊核估计 [J]. 电子学报,2017,45(5):1226-1233.

[5] PAN J S,HU Z,SU Z X,et al. Deblurring Text Images viaL0-Regularized Intensity and Gradient Prior [C]//2014 IEEE Conferenceon Computer Vision and Pattern Recognition.IEEE,2014:2901-2908.

[6] XU L,LU C W,XU Y,et al. Image smoothing via L0gradient minimization [J].ACM Transactions on Graphics,2011,30(6):1-12.

[7] CHO S Y,WANG J,LEE S Y. Handling outliers in non-blindimage deconvolution [C]//2011 International Conference on ComputerVision.IEEE,2011:495-502.

[8] LEVIN A,WEISS Y,DURAND F,et al. Efficient marginallikelihoodoptimization in blind deconvolution [C]//Proceedings of the2011 IEEE Conference on Computer Vision and Pattern Recognition.Washington:IEEE Computer Society,2011:2657-2664.

[9] LEVIN A,WEISS Y,DURAND F,et al. UnderstandingBlind Deconvolution Algorithms [J].IEEE Transactions on PatternAnalysis and Machine Intelligence,2011,33(12):2354-2367.


作者简介:

岳灿(1982—),男,汉族,湖北当阳人,助理工程师,硕士,研究方向:通信与信息系统;

官斌(1979—),男,汉族,湖北武汉人,工程师,硕士,研究方向:军事装备系统监造及效能评估;

杨栩萌(1988—),男,汉族,江苏连云港人,助理工程师,学士学位,研究方向:军用显控设备软硬件。