基于专家场先验模型的图像超分辨重建算法
|
摘要
为了进一步提高图像超分辨率重建的质量,针对非局部集中稀疏表示算法中重建图像的噪声问题,提出了一种基于专家场先验模型的图像超分辨率重建改进算法。首先,利用专家场模型从图像训练集中学习整幅图像的先验知识建立全局先验模型;然后将学习到的先验信息用于非局部集中稀疏表示模型求解最优稀疏表示系数;最后,得到高分辨率图像估计。该算法在超分辨率重建迭代运算的同时,同步更新专家场模型参数,因此在不显著增加运算复杂度的情况下,通过选取合适的先验约束,有效地增强了图像重建的效果。实验结果表明:相比非局部集中稀疏表示算法,文中算法对无噪和有噪降质图像均能取得较好的峰值信噪比结果,并且能够进一步提高有噪图像的去噪效果。
In order to further improve the quality of image Super-Resolution(SR) reconstruction, a SR reconstruction algorithm based on Fields of Experts(FoE) prior model was proposed for the noise problem of reconstructed image in Non-locally Centralized Sparse Representation(NCSR) algorithm.Firstly, the FoE model was used to learn the prior knowledge of the whole image from the image training data to establish the global prior model, and then the learned prior information was used to solve the optimal sparse representation coefficient in the NCSR framework. Finally, the high resolution image estimate was obtainon. The proposed algorithm updated parameters of FoE prior model while the SR reconstruction iterative operates. Therefore, the effect of image reconstruction can be effectively enhanced by selecting the appropriate prior constraints without significantly increasing the computational complexity.Compared with NCSR algorithm, the experimental results show that the proposed algorithm can obtain better peak signal to noise ratio results for both noiseless and noisy degradation images, and further improves the de-noising effect of noisy images.
In order to further improve the quality of image Super-Resolution(SR) reconstruction, a SR reconstruction algorithm based on Fields of Experts(FoE) prior model was proposed for the noise problem of reconstructed image in Non-locally Centralized Sparse Representation(NCSR) algorithm.Firstly, the FoE model was used to learn the prior knowledge of the whole image from the image training data to establish the global prior model, and then the learned prior information was used to solve the optimal sparse representation coefficient in the NCSR framework. Finally, the high resolution image estimate was obtainon. The proposed algorithm updated parameters of FoE prior model while the SR reconstruction iterative operates. Therefore, the effect of image reconstruction can be effectively enhanced by selecting the appropriate prior constraints without significantly increasing the computational complexity.Compared with NCSR algorithm, the experimental results show that the proposed algorithm can obtain better peak signal to noise ratio results for both noiseless and noisy degradation images, and further improves the de-noising effect of noisy images.
引文
[1] Park S C, Park M K, Kang M G. Super-resolution image reconstruction:A technical overview[J]. IEEE Signal Processing Magazine, 2003, 20(3):21-36.
[2] Boccacci P, Bertero A M. Introduction to Inverse Problems in Imaging[M]. Korea:Research Center for Peace and Unification of Korea, 1987:46-47.
[3] Elad M, Feuer A. Restoration of a single superresolution image from several blurred, noisy, and undersampled measured images[J]. IEEE Transactions on Image Processing A, 1997, 6(12):1646-1658.
[4] Irani M, Peleg S. Improving resolution by image registration[J]. Cvgip Graphical Models&Image Processing, 1991, 53(3):231-239.
[5] Baker S, Kanade T. Limits on super-resolution and how to break them[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24(9):0-1183.
[6] Freeman W T, Jones T R, Pasztor E C. Example-based super-resolution[J]. Computer Graphics&Applications IEEE, 2002, 22(2):56-65.
[7] Donoho D L. Compressed sensing[J]. IEEE Transactions on Information Theory, 2006, 52(4):1289-1306.
[8] Elad M, Aharon M. Image denoising via sparse and redundant representations over learned dictionaries[J]. IEEE Transactions on Image Processing, 2006, 15(12):3736-45.
[9] Dong W, Zhang L, Shi G. Centralized sparse representation for image restoration[C]//IEEE International Conference on Computer Vision, 2012.
[10] Dong W, Zhang L, Shi G, et al. Nonlocally centralized sparse representation for image restoration[J]. IEEE Transactions on Image Processing, 2013, 22(4):1620-1630.
[11] Mairal J, Bach F, Ponce J, et al. Non-local sparse models for image restoration[C]//IEEE, International Conference on Computer Vision, ICCV 2009, 2009:2272-2279.
[12] Ebrahimi M. Inverse problems and self-similarity in imaging[D]. Waterloo:University of Waterloo, 2008.
[13] Elad M, Datsenko D. Corrigendum:example-based regularization deployed to super-resolution reconstruction of a single image[J]. Computer Journal, 2009, 52(4):15-30.
[14] Li S Z. Markov Random Field Modeling in Iimage Analysis[M]. New York:Springer-Verlag, 2001.
[15] Roth S, Black M J. Fields of Experts:A framework for learning image priors[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005:860-867.
[16] Roth S, Black M J. Fields of experts[J]. International Journal of Computer Vision, 2009, 82(2):205-229.
[17] Welling M, Hinton G, Osindero S. Learning sparse topographic representations with products of student-t distribution[C]//Advances in Neural Information Processing Systems(NIPS′03), 2003:1359-1366.
[18] Dong W, Zhang L, Shi G, et al. Image deblurring and super-resolution by adaptive sparse domain selection and adaptive regularization[J]. IEEE Transactions on Image Processing, 2011, 20(7):2108306.
[2] Boccacci P, Bertero A M. Introduction to Inverse Problems in Imaging[M]. Korea:Research Center for Peace and Unification of Korea, 1987:46-47.
[3] Elad M, Feuer A. Restoration of a single superresolution image from several blurred, noisy, and undersampled measured images[J]. IEEE Transactions on Image Processing A, 1997, 6(12):1646-1658.
[4] Irani M, Peleg S. Improving resolution by image registration[J]. Cvgip Graphical Models&Image Processing, 1991, 53(3):231-239.
[5] Baker S, Kanade T. Limits on super-resolution and how to break them[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002, 24(9):0-1183.
[6] Freeman W T, Jones T R, Pasztor E C. Example-based super-resolution[J]. Computer Graphics&Applications IEEE, 2002, 22(2):56-65.
[7] Donoho D L. Compressed sensing[J]. IEEE Transactions on Information Theory, 2006, 52(4):1289-1306.
[8] Elad M, Aharon M. Image denoising via sparse and redundant representations over learned dictionaries[J]. IEEE Transactions on Image Processing, 2006, 15(12):3736-45.
[9] Dong W, Zhang L, Shi G. Centralized sparse representation for image restoration[C]//IEEE International Conference on Computer Vision, 2012.
[10] Dong W, Zhang L, Shi G, et al. Nonlocally centralized sparse representation for image restoration[J]. IEEE Transactions on Image Processing, 2013, 22(4):1620-1630.
[11] Mairal J, Bach F, Ponce J, et al. Non-local sparse models for image restoration[C]//IEEE, International Conference on Computer Vision, ICCV 2009, 2009:2272-2279.
[12] Ebrahimi M. Inverse problems and self-similarity in imaging[D]. Waterloo:University of Waterloo, 2008.
[13] Elad M, Datsenko D. Corrigendum:example-based regularization deployed to super-resolution reconstruction of a single image[J]. Computer Journal, 2009, 52(4):15-30.
[14] Li S Z. Markov Random Field Modeling in Iimage Analysis[M]. New York:Springer-Verlag, 2001.
[15] Roth S, Black M J. Fields of Experts:A framework for learning image priors[C]//IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2005:860-867.
[16] Roth S, Black M J. Fields of experts[J]. International Journal of Computer Vision, 2009, 82(2):205-229.
[17] Welling M, Hinton G, Osindero S. Learning sparse topographic representations with products of student-t distribution[C]//Advances in Neural Information Processing Systems(NIPS′03), 2003:1359-1366.
[18] Dong W, Zhang L, Shi G, et al. Image deblurring and super-resolution by adaptive sparse domain selection and adaptive regularization[J]. IEEE Transactions on Image Processing, 2011, 20(7):2108306.