A Logarithmic Image Prior for Blind Deconvolution

详细信息    查看全文

• 作者：Daniele Perrone ; Paolo Favaro
• 关键词：Blind deconvolution ; Majorization–minimization ; Primal ; dual ; Image prior ; Total variation ; Logarithmic prior
• 刊名：International Journal of Computer Vision
• 出版年：2016
• 出版时间：April 2016
• 年：2016
• 卷：117
• 期：2
• 页码：159-172
• 全文大小：3,049 KB
• 参考文献：Babacan, S. D., Molina, R., Do, M. N., & Katsaggelos, A. K. (2012). Bayesian blind deconvolution with general sparse image priors. In ECCV. Firenze: Springer.
  Burger, M., & Lucka, F. (2014). Maximum-a-posteriori estimates in linear inverse problems with log-concave priors are proper bayes estimators. Inverse Problems, 30, 114004.MathSciNet CrossRef MATH
  Candes, E. J., Wakin, M. B., & Boyd, S. (2008). Enhancing sparsity by reweighted l1 minimization. Journal of Fourier Analysis and Applications, 14(5), 877–905.MathSciNet CrossRef MATH
  Chambolle, A., & Pock, T. (2011). A first-order primal-dual algorithm for convex problems with applications to imaging. Journal of Mathematical Imaging and Vision, 40(1), 120–145.MathSciNet CrossRef MATH
  Chan, T., & Wong, C. K. (1998). Total variation blind deconvolution. IEEE Transactions on Image Processing, 7(3), 370–375.CrossRef
  Chaudhuri, S., Velmurugan, R., & Rameshan, R. M. (2014). Blind image deconvolution. Cham: Springer.CrossRef MATH
  Cho, S., & Lee, S. (2009). Fast motion deblurring. ACM Transactions on Graphics, 28(5), 1–8.CrossRef
  Fergus, R., Singh, B., Hertzmann, A., Roweis, S. T., & Freeman, W. T. (2006). Removing camera shake from a single photograph. ACM Transactions on Graphics, 25(3), 787–794.CrossRef
  Hunter, D., & Lange, K. (2004). A tutorial on mm algorithms. The American Statistician, 58, 30–37.MathSciNet CrossRef
  Kenig, T., Kam, Z., & Feuer, A. (2010). Blind image deconvolution using machine learning for three-dimensional microscopy. IEEE Transactions on Pattern Analysis and Machine Intelligence, 32(12), 2191–2204. doi:10.​1109/​TPAMI.​2010.​45 .CrossRef
  Keuper, M., Schmidt, T., Temerinac-Ott, M., Padeken, J., Heun, P., Ronneberger, O., & Brox, T. (2013). Blind deconvolution of widefield fluorescence microscopic data by regularization of the optical transfer function (otf). In 2013 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 2179–2186).
  Krishnan, D., Tay, T., & Fergus, R. (2011). Blind deconvolution using a normalized sparsity measure. In 2011 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 233–240).
  Krishnan, D., Bruna, J., & Fergus, R. (2013). Blind deconvolution with re-weighted sparsity promotion. CoRR arXiv:​1311.​4029 .
  Levin, A., Fergus, R., Durand, F., & Freeman, W. (2007). Image and depth from a conventional camera with a coded aperture. ACM Transactions on Graphics, 26, 70.CrossRef
  Levin, A., Weiss, Y., Durand, F., & Freeman, W. T. (2009). Understanding and evaluating blind deconvolution algorithms. In CVPR (pp. 1964–1971). IEEE.
  Levin, A., Weiss, Y., Durand, F., & Freeman, W. (2011a). Efficient marginal likelihood optimization in blind deconvolution. In 2011 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 2657–2664). doi:10.​1109/​CVPR.​2011.​5995308 .
  Levin, A., Weiss, Y., Durand, F., & Freeman, W. T. (2011b). Understanding blind deconvolution algorithms. IEEE Transactions on Pattern Analysis and Machine Intelligence, 33(12), 2354–2367.CrossRef
  Michaeli, T., & Irani, M. (2014). Blind deblurring using internal patch recurrence. In D. Fleet, T. Pajdla, B. Schiele, & T. Tuytelaars (Eds.) Computer Vision—ECCV 2014. Lecture Notes in Computer Science, vol. 8691 (pp. 783–798). Cham: Springer International Publishing. doi:10.​1007/​978-3-319-10578-9_​51 .
  Möllenhoff, T., Strekalovskiy, E., Möller, M., & Cremers, D. (2014a). Low rank priors for color image regularization. In Proceedings of 10th International Conference, Energy Minimization Methods in Computer Vision and Pattern Recognition, EMMCVPR 2015 (pp. 126–140). Hong Kong, China, January 13–16, 2015.
  Möllenhoff, T., Strekalovskiy, E., Möller, M., & Cremers, D. (2014b). The primal-dual hybrid gradient method for semiconvex splittings. CoRR arXiv:​1407.​1723 .
  Ochs, P., Chen, Y., Brox, T., & Pock, T. (2014). ipiano: Inertial proximal algorithm for nonconvex optimization. SIAM Journal on Imaging Sciences, 7(2), 1388–1419.MathSciNet CrossRef MATH
  Perrone, D., & Favaro, P. (2014). Total variation blind deconvolution: The devil is in the details. In The IEEE Conference on Computer Vision and Pattern Recognition (CVPR).
  Perrone, D., Diethelm, R., & Favaro, P. (2014). Blind deconvolution via lower-bounded logarithmic image priors. In Proceedings of 10th International Conference, Energy Minimization Methods in Computer Vision and Pattern Recognition, EMMCVPR 2015 (pp. 112–125). Hong Kong, China, January 13–16, 2015.
  Rockafellar, R. (1970). Convex analysis. Princeton, NJ: Princeton University Press.CrossRef MATH
  Rudin, L. I., Osher, S., & Fatemi, E. (1992). Nonlinear total variation based noise removal algorithms. Physica D, 60(1–4), 259–268. doi:10.​1016/​0167-2789(92)90242-F .MathSciNet CrossRef MATH
  Shan, Q., Jia, J., & Agarwala, A. (2008). High-quality motion deblurring from a single image. ACM Transactions on Graphics, 27, 73.
  Srivastava, A., Lee, A., Simoncelli, E. P., & Zhu, S. C. (2003). On advances in statistical modeling of natural images. Journal of Mathematical Imaging and Vision, 18, 17–33.MathSciNet CrossRef MATH
  Strekalovskiy, E., & Cremers, D. (2014). Real-time minimization of the piecewise smooth mumford-shah functional. Proceedings of Computer Vision 13th European Conference ECCV 2014, Zurich, Switzerland, September 6–12, 2014 Part II (pp. 127–141).
  Strong, D., & Chan, T. (2003). Edge-preserving and scale-dependent properties of total variation regularization. Inverse Problems, 19(6), S165.
  Sun, L., Cho, S., Wang, J., & Hays, J. (2013). Edge-based blur kernel estimation using patch priors. In 2013 IEEE International Conference on Computational Photography (ICCP) (pp. 1–8).
  Wipf, D., & Zhang, H. (2013). Analysis of Bayesian blind deconvolution. Energy minimization methods in computer vision and pattern recognition (pp. 40–53). Berlin: Springer.CrossRef
  Wipf, D., & Zhang, H. (2014). Revisiting bayesian blind deconvolution. Journal of Machine Learning Research, 15, 3595–3634.MathSciNet MATH
  Xu, L., & Jia, J. (2010). Two-phase kernel estimation for robust motion deblurring. In Proceedings of the 11th European Conference on Computer Vision: Part I ECCV’10 (pp. 157–170). Berlin: Springer.
  Xu, L., Zheng, S., & Jia, J. (2013). Unnatural l0 sparse representation for natural image deblurring. In 2013 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 1107–1114).
  Zoran, D., Weiss, Y. (2011). From learning models of natural image patches to whole image restoration. In Proceedings of the 2011 International Conference on Computer Vision, ICCV ’11 (pp. 479–486). IEEE Computer Society Washington, DC, USA.
  
• 作者单位：Daniele Perrone (1)
  Paolo Favaro (1)
  
  1. Department of Computer Science and Applied Mathematics, University of Bern, Neubrückstrasse 10, 3012, Bern, Switzerland
  
• 刊物类别：Computer Science
• 刊物主题：Computer Imaging, Vision, Pattern Recognition and Graphics
  Artificial Intelligence and Robotics
  Image Processing and Computer Vision
  Pattern Recognition
  
• 出版者：Springer Netherlands
• ISSN：1573-1405

文摘

Blind Deconvolution consists in the estimation of a sharp image and a blur kernel from an observed blurry image. Because the blur model admits several solutions it is necessary to devise an image prior that favors the true blur kernel and sharp image. Many successful image priors enforce the sparsity of the sharp image gradients. Ideally the \(L_0\) “norm” is the best choice for promoting sparsity, but because it is computationally intractable, some methods have used a logarithmic approximation. In this work we also study a logarithmic image prior. We show empirically how well the prior suits the blind deconvolution problem. Our analysis confirms experimentally the hypothesis that a prior should not necessarily model natural image statistics to correctly estimate the blur kernel. Furthermore, we show that a simple Maximum a Posteriori formulation is enough to achieve state of the art results. To minimize such formulation we devise two iterative minimization algorithms that cope with the non-convexity of the logarithmic prior: one obtained via the primal-dual approach and one via majorization-minimization.