AI helps you reading Science

AI generates interpretation videos

AI extracts and analyses the key points of the paper to generate videos automatically


pub
Go Generating

AI Traceability

AI parses the academic lineage of this thesis


Master Reading Tree
Generate MRT

AI Insight

AI extracts a summary of this paper


Weibo:
We showed that contrary to the common belief, the Point Spread Function of the camera is the wrong blur kernel k to use in SR algorithms

Nonparametric Blind Super-resolution

Computer Vision, no. 1 (2013): 945-952

Cited: 201|Views58
EI WOS

Abstract

Super resolution (SR) algorithms typically assume that the blur kernel is known (either the Point Spread Function 'PSF' of the camera, or some default low-pass filter, e.g. a Gaussian). However, the performance of SR methods significantly deteriorates when the assumed blur kernel deviates from the true one. We propose a general framework ...More

Code:

Data:

0
Introduction
  • Super-resolution (SR) refers to the task of recovering a high-resolution image from one or several of its lowresolution versions.
  • [11, 7, 2, 4, 20, 16, 15, 21]) assume that the low-resolution input image was obtained by down-sampling the unknown high-resolution image with a known blur kernel.
  • The authors show that unlike the common belief, even if the PSF is known, it is the wrong blur kernel to use in SR algorithms! The authors further show how to obtain the optimal SR blur kernel directly from the low-resolution image
Highlights
  • Super-resolution (SR) refers to the task of recovering a high-resolution image from one or several of its lowresolution versions
  • Relying on the wrong blur kernel may lead to low-quality SR results, as demonstrated in Fig. 1
  • We show that unlike the common belief, even if the Point Spread Function (PSF) is known, it is the wrong blur kernel to use in SR algorithms! We further show how to obtain the optimal SR blur kernel directly from the low-resolution image
  • We use the method of [8] as a representative of SR methods that rely on internal patch recurrence, and the algorithm of [21] as representative of SR methods that train on an external database of examples
  • For the external kernel recovery we used a database of 30 natural images downloaded from the Internet
  • We showed that contrary to the common belief, the PSF of the camera is the wrong blur kernel k to use in SR algorithms
Results
  • The authors validated the benefit of using the kernel estimation in SR algorithms both empirically, as well as visually on real images.
  • The authors use the method of [8] as a representative of SR methods that rely on internal patch recurrence, and the algorithm of [21] as representative of SR methods that train on an external database of examples.
  • In the experiments the low-res patches q and rα were 5 × 5 patches.
  • The authors experimented with SR×2 and SR×3.
  • For the external kernel recovery the authors used a database of 30 natural images downloaded from the Internet.
  • The regularization in the least-squares stage of each iteration was performed with a matrix C corresponding to derivatives in the x and y directions
Conclusion
  • The authors showed that contrary to the common belief, the PSF of the camera is the wrong blur kernel k to use in SR algorithms.
  • The authors showed that the correct k can be recovered directly from the low-res image, by exploiting the recur-.
  • Rence of small image patches.
  • This is shown to be a principled MAP estimation of k, and leads to a significant improvement in SR results
Reference
  • I. Begin and F. R. Ferrie. Blind super-resolution using a learning-based approach. In ICPR, 2004.
    Google ScholarLocate open access versionFindings
  • H. Chang, D.-Y. Yeung, and Y. Xiong. Super-resolution through neighbor embedding. In CVPR, 2004.
    Google ScholarLocate open access versionFindings
  • Y. C. Eldar and T. Michaeli. Beyond bandlimited sampling. IEEE Signal Process. Mag., 26(3):48–68, 2009.
    Google ScholarLocate open access versionFindings
  • R. Fattal. Image upsampling via imposed edge statistics. ACM TOG, 26(3):95, 2007.
    Google ScholarLocate open access versionFindings
  • R. Fergus, B. Singh, A. Hertzmann, S. T. Roweis, and W. T. Freeman. Removing camera shake from a single photograph. ACM TOG, 25(3):787–794, 2006.
    Google ScholarLocate open access versionFindings
  • G. Freedman and R. Fattal. Image and video upscaling from local self-examples. ACM TOG, 30(2):12, 2011.
    Google ScholarLocate open access versionFindings
  • W. T. Freeman, T. R. Jones, and E. C. Pasztor. Examplebased super-resolution. IEEE Computer Graphics and Applications, 22(2):56–65, 2002.
    Google ScholarLocate open access versionFindings
  • D. Glasner, S. Bagon, and M. Irani. Super-resolution from a single image. In ICCV, 2009.
    Google ScholarLocate open access versionFindings
  • S. Harmeling, S. Sra, M. Hirsch, and B. Scholkopf. Multiframe blind deconvolution, super-resolution, and saturation correction via incremental EM. In ICIP, 2010.
    Google ScholarLocate open access versionFindings
  • Y. He, K. H. Yap, L. Chen, and L. P. Chau. A soft MAP framework for blind super-resolution image reconstruction. Image and Vision Computing, 27(4):364 – 373, 2009.
    Google ScholarLocate open access versionFindings
  • M. Irani and S. Peleg. Improving resolution by image registration. CVGIP, 53(3):231–239, 1991.
    Google ScholarLocate open access versionFindings
  • N. Joshi, R. Szeliski, and D. J. Kriegman. PSF estimation using sharp edge prediction. In CVPR, 2008.
    Google ScholarLocate open access versionFindings
  • A. Levin, Y. Weiss, F. Durand, and W. T. Freeman. Understanding and evaluating blind deconvolution algorithms. In CVPR, 2009.
    Google ScholarLocate open access versionFindings
  • A. Levin, Y. Weiss, F. Durand, and W. T. Freeman. Efficient marginal likelihood optimization in blind deconvolution. In CVPR, 2011.
    Google ScholarLocate open access versionFindings
  • S. Mallat and G. Yu. Super-resolution with sparse mixing estimators. Trans. Signal Proc., 19(11):2889–2900, 2010.
    Google ScholarLocate open access versionFindings
  • P. Milanfar, editor. Super-resolution imaging. CRC Press, 2010.
    Google ScholarFindings
  • J. Qiao, J. Liu, and C. Zhao. A novel SVM-based blind super-resolution algorithm. In International Joint Conference on Neural Networks, 2006.
    Google ScholarLocate open access versionFindings
  • F. Sroubek, G. Cristobal, and J. Flusser. Simultaneous superresolution and blind deconvolution. In Journal of Physics: Conference Series, volume 124, 2008.
    Google ScholarLocate open access versionFindings
  • Q. Wang, X. Tang, and H. Shum. Patch based blind image super resolution. In ICCV, 2005.
    Google ScholarLocate open access versionFindings
  • J. Yang, J. Wright, T. Huang, and Y. Ma. Image superresolution as sparse representation of raw image patches. In CVPR, 2008.
    Google ScholarLocate open access versionFindings
  • R. Zeyde, M. Elad, and M. Protter. On single image scale-up using sparse-representations. In Curves and Surfaces, 2012.
    Google ScholarLocate open access versionFindings
  • M. Zontak and M. Irani. Internal statistics of a single natural image. In CVPR, 2011.
    Google ScholarLocate open access versionFindings
0
Your rating :

No Ratings

Tags
Comments
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn