American Institute of Mathematical Sciences

Advanced
June  2021, 15(3): 475-498. doi: 10.3934/ipi.2021001

Tensor train rank minimization with nonlocal self-similarity for tensor completion

Meng Ding 1, , Ting-Zhu Huang 1,, , Xi-Le Zhao 1,, , Michael K. Ng 2, and  Tian-Hui Ma 3,

1. 

School of Mathematical Sciences, University of Electronic Science and Technology of China, Chengdu, 611731, China
2. 

Department of Mathematics, The University of Hong Kong, Pokfulam, Hong Kong
3. 

School of Mathematics and Statistics, Xi'an Jiaotong University, Xi'an, 710049, China

* Corresponding authors: Ting-Zhu Huang and Xi-Le Zhao

Received  February 2020 Revised  September 2020 Published  December 2020

The tensor train (TT) rank has received increasing attention in tensor completion due to its ability to capture the global correlation of high-order tensors ($ \rm{order} >3 $). For third order visual data, direct TT rank minimization has not exploited the potential of TT rank for high-order tensors. The TT rank minimization accompany with ket augmentation, which transforms a lower-order tensor (e.g., visual data) into a higher-order tensor, suffers from serious block-artifacts. To tackle this issue, we suggest the TT rank minimization with nonlocal self-similarity for tensor completion by simultaneously exploring the spatial, temporal/spectral, and nonlocal redundancy in visual data. More precisely, the TT rank minimization is performed on a formed higher-order tensor called group by stacking similar cubes, which naturally and fully takes advantage of the ability of TT rank for high-order tensors. Moreover, the perturbation analysis for the TT low-rankness of each group is established. We develop the alternating direction method of multipliers tailored for the specific structure to solve the proposed model. Extensive experiments demonstrate that the proposed method is superior to several existing state-of-the-art methods in terms of both qualitative and quantitative measures.

Keywords: Low-rank tensor completion, tensor train rank, nonlocal self-similarity, alternating direction method of multipliers.
Mathematics Subject Classification: Primary: 94A08; Secondary: 68U10.
Citation: Meng Ding, Ting-Zhu Huang, Xi-Le Zhao, Michael K. Ng, Tian-Hui Ma. Tensor train rank minimization with nonlocal self-similarity for tensor completion. Inverse Problems & Imaging, 2021, 15 (3) : 475-498. doi: 10.3934/ipi.2021001
References:
[1]

J. A. Bengua, H. N. Phiem, H. D. Tuan and M. N. Do, Efficient tensor completion for color image and video recovery: Low-rank tensor train, IEEE Transactions on Image Processing, 26 (2017), 2466-2479.  doi: 10.1109/TIP.2017.2672439.  Google Scholar

[2]

D. P. Bertsekas, A. Nedic and A. E. Ozdaglar, Convex Analysis and Optimization, Athena Scientific, 2003.  Google Scholar

[3]

M. Bertalmio, G. Sapiro, V. Caselles and C. Ballester, Image inpainting, Siggraph, 4 (2000), 417-424.  doi: 10.21236/ADA437378.  Google Scholar

[4]

J.-F. Cai, E. J. Cand$\grave{e}$s and Z. Shen, A singular value thresholding algorithm for matrix completion, SIAM Journal on Optimization, 20 (2010), 1956-1982.  doi: 10.1137/080738970.  Google Scholar

[5]

S. H. Chan, R. Khoshabeh, K. B. Gibson, P. E. Gill and T. Q. Nguyen, An augmented Lagrangian method for total variation video restoration, IEEE Transactions on Image Processing, 20 (2011), 3097-3111.  doi: 10.1109/TIP.2011.2158229.  Google Scholar

[6]

R. H. Chan, M. Tao and X. Yuan, Constrained total variation deblurring models and fast algorithms based on alternating direction method of multipliers, SIAM Journal on Imaging Sciences, 6 (2013), 680-697.  doi: 10.1137/110860185.  Google Scholar

[7]

Y. Chang, L.-X. Yan and S. Zhong, Hyper-laplacian regularized unidirectional low-rank tensor recovery for multispectral image denoising, IEEE Conference on Computer Vision and Pattern Recognition, (2017), 5901–5909. doi: 10.1109/CVPR.2017.625.  Google Scholar

[8]

Y. Chen, C. Hsu and H. M. Liao, Simultaneous tensor decomposition and completion using factor priors, IEEE Transactions on Pattern Analysis and Machine Intelligence, 20 (2014), 577-591.   Google Scholar

[9]

L.-B. Cui, X.-Q. Zhang and S.-L. Wu, A new preconditioner of the tensor splitting iterative method for solving multi-linear systems with $\mathcal{M}$-tensors, Computational and Applied Mathematics, 39 (2020), 1-16.  doi: 10.1007/s40314-020-01194-8.  Google Scholar

[10]

K. Dabov, A. Foi, V. Katkovnik and K. Egiazarian, Image denoising by sparse 3-D transform-domain collaborative filtering, IEEE Transactions on Image Processing, 16 (2007), 2080-2095.  doi: 10.1109/TIP.2007.901238.  Google Scholar

[11]

M. Ding, T.-Z. Huang and T.-H. Ma, Cauchy noise removal using group-based low-rank prior, Applied Mathematics and Computation, 372 (2020), 124971, 15 pp. doi: 10.1016/j.amc.2019.124971.  Google Scholar

[12]

M. Ding, T.-Z. Huang, T.-Y. Ji, X.-L. Zhao and J.-H. Yang, Low-rank tensor completion using matrix factorization based on tensor train rank and total variation, Journal of Scientific Computing, 81 (2019), 941-964.  doi: 10.1007/s10915-019-01044-8.  Google Scholar

[13]

M. Ding, T.-Z. Huang, S. Wang, J.-J. Mei and X.-L. Zhao, Total variation with overlapping group sparsity for deblurring images under Cauchy noise, Applied Mathematics and Computation, 341 (2019), 128-147.  doi: 10.1016/j.amc.2018.08.014.  Google Scholar

[14]

Y. Du, G. Han, Y. Quan, Z. Yu, H. Wong, C. L. P. Chen and J. Zhang, Exploiting global low-rank structure and local sparsity nature for tensor completion, IEEE Transactions on Cybernetics, 49 (2019), 3898-3910.  doi: 10.1109/TCYB.2018.2853122.  Google Scholar

[15]

J. Eckstein and D. P. Bertsekas, On the Douglas-Rachford splitting method and the proximal point algorithm for maximal monotone operators, Mathematical Programming, 55 (1992), 293-318.  doi: 10.1007/BF01581204.  Google Scholar

[16]

G. Ely, S. Aeron, N. Hao and M. E. Kilmer, 5D seismic data completion and denoising using a novel class of tensor decompositions, GEOPHYSICS, 80 (2015), V83–V95. doi: 10.1190/geo2014-0467.1.  Google Scholar

[17]

X. Fu, K.-J. Huang, B. Yang, W. K. Ma and N. D. Sidiropoulos, Robust volume minimization-based matrix factorization for remote sensing and document clustering, IEEE Transactions on Signal Processing, 64 (2016), 6254-6268.  doi: 10.1109/TSP.2016.2602800.  Google Scholar

[18]

S. Gandy, B. Recht and I. Yamada, Tensor completion and low-n-rank tensor recovery via convex optimization, Inverse Problems, 27 (2011), 025010, 19pp. doi: 10.1088/0266-5611/27/2/025010.  Google Scholar

[19]

T. Goldstein, B. O'Donoghue, S. Setzer and R. Baraniuk, Fast alternating direction optimization methods, SIAM Journal on Imaging Sciences, 7 (2014), 1588-1623.  doi: 10.1137/120896219.  Google Scholar

[20]

L. Grasedyck, M. Kluge and S. Krämer, Alternating least squares tensor completion in the TT-format, preprint, arXiv: 1509.00311. Google Scholar

[21]

S.-H. Gu, L. Zhang, W.-M. Zuo and X.-C. Feng, Weighted nuclear norm minimization with application to image denoising, IEEE Conference on Computer Vision and Pattern Recognition, (2014), 2862–2869. doi: 10.1109/CVPR.2014.366.  Google Scholar

[22]

B.-S. He and X. Yuan, On the O(1/n) convergence rate of the douglas-rachford alternating direction method, SIAM Journal on Numerical Analysis, 50 (2012), 700-709.  doi: 10.1137/110836936.  Google Scholar

[23]

W. He, H.-Y. Zhang, L.-P. Zhang and H.-F. Shen, Total-variation-regularized low-rank matrix factorization for hyperspectral image restoration, IEEE Transactions on Geoscience and Remote Sensing, 54 (2016), 178-188.  doi: 10.1109/TGRS.2015.2452812.  Google Scholar

[24]

C. J. Hillar and L. H. Lim, Most tensor problems are NP-hard, Journal of the ACM, 60 (2013), Art. 45, 39 pp. doi: 10.1145/2512329.  Google Scholar

[25]

Y.-M. Huang, H.-Y. Yan, Y.-W. Wen and X. Yang, Rank minimization with applications to image noise removal, Information Sciences, 429 (2018), 147-163.  doi: 10.1016/j.ins.2017.10.047.  Google Scholar

[26]

T.-X. Jiang, T.-Z. Huang, X.-L. Zhao and L.-J. Deng, Multi-dimensional imaging data recovery via minimizing the partial sum of tubal nuclear norm, Journal of Computational and Applied Mathematics, 372 (2020), 112680, 15pp. doi: 10.1016/j.cam.2019.112680.  Google Scholar

[27]

T.-X. Jiang, M. K. Ng, X.-L. Zhao and T.-Z. Huang, Framelet representation of tensor nuclear norm for third-order tensor completion, IEEE Transactions on Image Processing, 29 (2020), 7233-7244.  doi: 10.1109/TIP.2020.3000349.  Google Scholar

[28]

T. G. Kolda, B. W. Bader and J. P. Kenny, Higher-order Web link analysis using multilinear algebra, IEEE International Conference on Data Mining, (2005), 242–249. doi: 10.1109/ICDM.2005.77.  Google Scholar

[29]

T. G. Kolda and B. W. Bader, Tensor decompositions and applications, SIAM Review, 51 (2009), 455-500.  doi: 10.1137/07070111X.  Google Scholar

[30]

M. E. Kilmer, K. Braman, N. Hao and R. C. Hoover, Third-order tensors as operators on matrices: A theoretical and computational framework with applications in imaging, SIAM Journal on Matrix Analysis and Applications, 34 (2013), 148-172.  doi: 10.1137/110837711.  Google Scholar

[31]

N. Komodakis, Image inpainting, IEEE Conference on Computer Vision and Pattern Recognition, 1 (2006), 442-452.   Google Scholar

[32]

R.-J. Lai and J. Li, Manifold based low-rank regularization for image restoration and semi-supervised learning, Journal of Scientific Computing, 74 (2018), 1241-1263.  doi: 10.1007/s10915-017-0492-x.  Google Scholar

[33]

J. I. Latorre, Image Compression and Entanglement, Computer Science, 2005. Google Scholar

[34]

F. Li, M. K. Ng and R. J. Plemmons, Coupled segmentation and denoising/deblurring models for hyperspectral material identification, Numerical Linear Algebra with Applications, 19 (2012), 153-173.  doi: 10.1002/nla.750.  Google Scholar

[35]

Y.-P. Liu, Z. Long and C. Zhu, Image completion using low tensor tree rank and total variation minimization, IEEE Transactions on Multimedia, 21 (2019), 338-350.  doi: 10.1109/TMM.2018.2859026.  Google Scholar

[36]

Y.-Y. Liu, F.-H. Shang, L.-C. Jiao, J. Cheng and H. Cheng, Trace norm regularized CANDECOMP/PARAFAC decomposition with missing data, IEEE Transactions on Cybernetics, 45 (2015), 2437-2448.  doi: 10.1109/TCYB.2014.2374695.  Google Scholar

[37]

Y.-P. Liu, Z. Long, H.-Y. Huang and C. Zhu, Low CP rank and tucker rank tensor completion for estimating missing components in image data, IEEE Transactions on Circuits and Systems for Video Technology, 30 (2020), 944-954.  doi: 10.1109/TCSVT.2019.2901311.  Google Scholar

[38]

J. Liu, P. Musialski, P. Wonka and J. Ye, Tensor completion for estimating missing values in visual data, IEEE Transactions on Pattern Analysis and Machine Intelligence, 35 (2013), 208-220.  doi: 10.1109/TPAMI.2012.39.  Google Scholar

[39]

C.-Y. Lu, J.-S. Feng, Y.-D. Chen, W. Liu, Z.-C. Lin and S.-C. Yan, Tensor robust principal component analysis: Exact recovery of corrupted low-rank tensors via convex optimization, IEEE Conference on Computer Vision and Pattern Recognition, (2016), 5249–5257. doi: 10.1109/CVPR.2016.567.  Google Scholar

[40]

C.-Y. Lu, J.-S. Feng, Z.-C. Lin and S.-C. Yan, Exact Low Tubal Rank Tensor Recovery from Gaussian Measurements, International Joint Conference on Artificial Intelligence, 2018. doi: 10.24963/ijcai.2018/347.  Google Scholar

[41]

I. V. Oseledets, Tensor-train decomposition, SIAM Journal on Scientific Computing, 33 (2011), 2295-2317.  doi: 10.1137/090752286.  Google Scholar

[42]

S. Osher, Z.-Q. Shi and W. Zhu, Low dimensional manifold model for image processing, SIAM Journal on Imaging Sciences, 10 (2017), 1669-1690.  doi: 10.1137/16M1058686.  Google Scholar

[43]

E. E. Papalexakis, C. Faloutsos and N. D. Sidiropoulos, Tensors for data mining and data fusion: Models, applications, and scalable algorithms, ACM Transactions on Intelligent Systems and Technology, 8 (2017), 16: 1–16: 44. doi: 10.1145/2915921.  Google Scholar

[44]

Z.-Q. Shi, S. Osher and W. Zhu, Weighted nonlocal laplacian on interpolation from sparse data, Journal of Scientific Computing, 73 (2017), 1164-1177.  doi: 10.1007/s10915-017-0421-z.  Google Scholar

[45]

N. D. Sidiropoulos, L. De Lathauwer, X. Fu, K. Huang, E. E. Papalexakis and C. Faloutsos, Tensor decomposition for signal processing and machine learning, IEEE Transactions on Signal Processing, 65 (2017), 3551-3582.  doi: 10.1109/TSP.2017.2690524.  Google Scholar

[46]

G. W. Stewart, Matrix Algorithms, Vol I: Basic Decomposition, SIAM, 2001. doi: 10.1137/1.9780898718058.  Google Scholar

[47]

W. Wang, V. Aggarwal and S. Aeron, Tensor completion by alternating minimization under the tensor train (TT) model, preprint, arXiv: 1609.05587. Google Scholar

[48]

Y. Wang, J.-J. Peng, Q. Zhao, Y. Leung, X.-L. Zhao and D.-Y. Meng, Hyperspectral image restoration via total variation regularized low-rank tensor decomposition, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 11 (2018), 1227-1243.   Google Scholar

[49]

Y. Wang, D.-Y. Meng and M. Yuan, Sparse recovery: From vectors to tensors, National Science Review, 5 (2018), 756-767.  doi: 10.1093/nsr/nwx069.  Google Scholar

[50]

T. Xie, S.-T. Li, L.-Y. Fang and L.-C. Liu, Tensor completion via nonlocal low-rank regularization, IEEE Transactions on Cybernetics, 49 (2019), 2344-2354.  doi: 10.1109/TCYB.2018.2825598.  Google Scholar

[51]

Z.-M. Xing, M.-Y. Zhou, A. Castrodad, G. Sapiro and L. Carin, Dictionary learning for noisy and incomplete hyperspectral images, SIAM Journal on Imaging Sciences, 5 (2012), 33-56.  doi: 10.1137/110837486.  Google Scholar

[52]

B. Xiong, Q.-G. Liu, J.-J. Xiong, S.-Q. Li, S.-S. Wang and D. Liang, Field-of-experts filters guided tensor completion, IEEE Transactions on Multimedia, 20 (2018), 2316-2329.  doi: 10.1109/TMM.2018.2806225.  Google Scholar

[53]

R.-T. Xu, Y. Xu and Y.-H. Quan, Factorized tensor dictionary learning for visual tensor data completion, IEEE Transactions on Multimedia, (2020), 1–14. Google Scholar

[54]

Y.-Y. Xu, R.-R. Hao, W.-T. Yin and Z.-X. Su, Parallel matrix factorization for low-rank tensor completion, Inverse Problems and Imaging, 9 (2017), 601-624.  doi: 10.3934/ipi.2015.9.601.  Google Scholar

[55]

J.-H. Yang, X.-L. Zhao, T.-H. Ma, M. Ding and T.-Z. Huang, Tensor train rank minimization with hybrid smoothness for visual data recovery, Applied Mathematical Modelling, 81 (2020), 711-726.  doi: 10.1016/j.apm.2020.01.039.  Google Scholar

[56]

J.-H. Yang, X.-L. Zhao, T.-H. Ma, Y. Chen, T.-Z. Huang and M. Ding, Remote sensing image destriping using unidirectional hybrid total variation and nonconvex low-rank regularization, Journal of Computational and Applied Mathematics, 363 (2020), 124-144.  doi: 10.1016/j.cam.2019.06.004.  Google Scholar

[57]

T. Yokota, Q.-B. Zhao, C. Li and A. Cichocki, Smooth PARAFAC decomposition for tensor completion, IEEE Transactions on Signal Processing, 64 (2016), 5423-5436.  doi: 10.1109/TSP.2016.2586759.  Google Scholar

[58]

Q.-B. Zhao, L-Q. Zhang and A. Cichocki, Bayesian CP factorization of incomplete tensors with automatic rank determination, IEEE Transactions on Pattern Analysis and Machine Intelligence, 37 (2015), 1751-1763.  doi: 10.1109/TPAMI.2015.2392756.  Google Scholar

[59]

X.-L. Zhao, W. Wang, T.-Y. Zeng, T.-Z. Huang and M. K. Ng, Total variation structured total least squares method for image restoration, SIAM Journal on Scientific Computing, 35 (2013), 1304-1320.  doi: 10.1137/130915406.  Google Scholar

[60]

X.-J. Zhang, A nonconvex relaxation approach to low-rank tensor completion, IEEE Transactions on Neural Networks and Learning Systems, 30 (2019), 1659-1671.  doi: 10.1109/TNNLS.2018.2872583.  Google Scholar

[61]

K.-B. Zhang, X.-B. Gao, D. -C.Tao and X.-L. Li, Single image super-resolution with non-local means and steering kernel regression, IEEE Transactions on Image Processing, 21 (2012), 4544-4556.  doi: 10.1109/TIP.2012.2208977.  Google Scholar

[62]

Z. Zhang, G. Ely and S. Aeron, Exact tensor completion using t-SVD, IEEE Transactions on Signal Processing, 65 (2017), 1511-1526.  doi: 10.1109/TSP.2016.2639466.  Google Scholar

[63]

Y.-B. Zheng, T.-Z. Huang, X.-L. Zhao, Y. Chen and W. He, Double-factor-regularized low-rank tensor factorization for mixed noise removal in hyperspectral image, IEEE Transactions on Geoscience and Remote Sensing, (2020), 1–15. Google Scholar

show all references

References:
[1]

J. A. Bengua, H. N. Phiem, H. D. Tuan and M. N. Do, Efficient tensor completion for color image and video recovery: Low-rank tensor train, IEEE Transactions on Image Processing, 26 (2017), 2466-2479.  doi: 10.1109/TIP.2017.2672439.  Google Scholar

[2]

D. P. Bertsekas, A. Nedic and A. E. Ozdaglar, Convex Analysis and Optimization, Athena Scientific, 2003.  Google Scholar

[3]

M. Bertalmio, G. Sapiro, V. Caselles and C. Ballester, Image inpainting, Siggraph, 4 (2000), 417-424.  doi: 10.21236/ADA437378.  Google Scholar

[4]

J.-F. Cai, E. J. Cand$\grave{e}$s and Z. Shen, A singular value thresholding algorithm for matrix completion, SIAM Journal on Optimization, 20 (2010), 1956-1982.  doi: 10.1137/080738970.  Google Scholar

[5]

S. H. Chan, R. Khoshabeh, K. B. Gibson, P. E. Gill and T. Q. Nguyen, An augmented Lagrangian method for total variation video restoration, IEEE Transactions on Image Processing, 20 (2011), 3097-3111.  doi: 10.1109/TIP.2011.2158229.  Google Scholar

[6]

R. H. Chan, M. Tao and X. Yuan, Constrained total variation deblurring models and fast algorithms based on alternating direction method of multipliers, SIAM Journal on Imaging Sciences, 6 (2013), 680-697.  doi: 10.1137/110860185.  Google Scholar

[7]

Y. Chang, L.-X. Yan and S. Zhong, Hyper-laplacian regularized unidirectional low-rank tensor recovery for multispectral image denoising, IEEE Conference on Computer Vision and Pattern Recognition, (2017), 5901–5909. doi: 10.1109/CVPR.2017.625.  Google Scholar

[8]

Y. Chen, C. Hsu and H. M. Liao, Simultaneous tensor decomposition and completion using factor priors, IEEE Transactions on Pattern Analysis and Machine Intelligence, 20 (2014), 577-591.   Google Scholar

[9]

L.-B. Cui, X.-Q. Zhang and S.-L. Wu, A new preconditioner of the tensor splitting iterative method for solving multi-linear systems with $\mathcal{M}$-tensors, Computational and Applied Mathematics, 39 (2020), 1-16.  doi: 10.1007/s40314-020-01194-8.  Google Scholar

[10]

K. Dabov, A. Foi, V. Katkovnik and K. Egiazarian, Image denoising by sparse 3-D transform-domain collaborative filtering, IEEE Transactions on Image Processing, 16 (2007), 2080-2095.  doi: 10.1109/TIP.2007.901238.  Google Scholar

[11]

M. Ding, T.-Z. Huang and T.-H. Ma, Cauchy noise removal using group-based low-rank prior, Applied Mathematics and Computation, 372 (2020), 124971, 15 pp. doi: 10.1016/j.amc.2019.124971.  Google Scholar

[12]

M. Ding, T.-Z. Huang, T.-Y. Ji, X.-L. Zhao and J.-H. Yang, Low-rank tensor completion using matrix factorization based on tensor train rank and total variation, Journal of Scientific Computing, 81 (2019), 941-964.  doi: 10.1007/s10915-019-01044-8.  Google Scholar

[13]

M. Ding, T.-Z. Huang, S. Wang, J.-J. Mei and X.-L. Zhao, Total variation with overlapping group sparsity for deblurring images under Cauchy noise, Applied Mathematics and Computation, 341 (2019), 128-147.  doi: 10.1016/j.amc.2018.08.014.  Google Scholar

[14]

Y. Du, G. Han, Y. Quan, Z. Yu, H. Wong, C. L. P. Chen and J. Zhang, Exploiting global low-rank structure and local sparsity nature for tensor completion, IEEE Transactions on Cybernetics, 49 (2019), 3898-3910.  doi: 10.1109/TCYB.2018.2853122.  Google Scholar

[15]

J. Eckstein and D. P. Bertsekas, On the Douglas-Rachford splitting method and the proximal point algorithm for maximal monotone operators, Mathematical Programming, 55 (1992), 293-318.  doi: 10.1007/BF01581204.  Google Scholar

[16]

G. Ely, S. Aeron, N. Hao and M. E. Kilmer, 5D seismic data completion and denoising using a novel class of tensor decompositions, GEOPHYSICS, 80 (2015), V83–V95. doi: 10.1190/geo2014-0467.1.  Google Scholar

[17]

X. Fu, K.-J. Huang, B. Yang, W. K. Ma and N. D. Sidiropoulos, Robust volume minimization-based matrix factorization for remote sensing and document clustering, IEEE Transactions on Signal Processing, 64 (2016), 6254-6268.  doi: 10.1109/TSP.2016.2602800.  Google Scholar

[18]

S. Gandy, B. Recht and I. Yamada, Tensor completion and low-n-rank tensor recovery via convex optimization, Inverse Problems, 27 (2011), 025010, 19pp. doi: 10.1088/0266-5611/27/2/025010.  Google Scholar

[19]

T. Goldstein, B. O'Donoghue, S. Setzer and R. Baraniuk, Fast alternating direction optimization methods, SIAM Journal on Imaging Sciences, 7 (2014), 1588-1623.  doi: 10.1137/120896219.  Google Scholar

[20]

L. Grasedyck, M. Kluge and S. Krämer, Alternating least squares tensor completion in the TT-format, preprint, arXiv: 1509.00311. Google Scholar

[21]

S.-H. Gu, L. Zhang, W.-M. Zuo and X.-C. Feng, Weighted nuclear norm minimization with application to image denoising, IEEE Conference on Computer Vision and Pattern Recognition, (2014), 2862–2869. doi: 10.1109/CVPR.2014.366.  Google Scholar

[22]

B.-S. He and X. Yuan, On the O(1/n) convergence rate of the douglas-rachford alternating direction method, SIAM Journal on Numerical Analysis, 50 (2012), 700-709.  doi: 10.1137/110836936.  Google Scholar

[23]

W. He, H.-Y. Zhang, L.-P. Zhang and H.-F. Shen, Total-variation-regularized low-rank matrix factorization for hyperspectral image restoration, IEEE Transactions on Geoscience and Remote Sensing, 54 (2016), 178-188.  doi: 10.1109/TGRS.2015.2452812.  Google Scholar

[24]

C. J. Hillar and L. H. Lim, Most tensor problems are NP-hard, Journal of the ACM, 60 (2013), Art. 45, 39 pp. doi: 10.1145/2512329.  Google Scholar

[25]

Y.-M. Huang, H.-Y. Yan, Y.-W. Wen and X. Yang, Rank minimization with applications to image noise removal, Information Sciences, 429 (2018), 147-163.  doi: 10.1016/j.ins.2017.10.047.  Google Scholar

[26]

T.-X. Jiang, T.-Z. Huang, X.-L. Zhao and L.-J. Deng, Multi-dimensional imaging data recovery via minimizing the partial sum of tubal nuclear norm, Journal of Computational and Applied Mathematics, 372 (2020), 112680, 15pp. doi: 10.1016/j.cam.2019.112680.  Google Scholar

[27]

T.-X. Jiang, M. K. Ng, X.-L. Zhao and T.-Z. Huang, Framelet representation of tensor nuclear norm for third-order tensor completion, IEEE Transactions on Image Processing, 29 (2020), 7233-7244.  doi: 10.1109/TIP.2020.3000349.  Google Scholar

[28]

T. G. Kolda, B. W. Bader and J. P. Kenny, Higher-order Web link analysis using multilinear algebra, IEEE International Conference on Data Mining, (2005), 242–249. doi: 10.1109/ICDM.2005.77.  Google Scholar

[29]

T. G. Kolda and B. W. Bader, Tensor decompositions and applications, SIAM Review, 51 (2009), 455-500.  doi: 10.1137/07070111X.  Google Scholar

[30]

M. E. Kilmer, K. Braman, N. Hao and R. C. Hoover, Third-order tensors as operators on matrices: A theoretical and computational framework with applications in imaging, SIAM Journal on Matrix Analysis and Applications, 34 (2013), 148-172.  doi: 10.1137/110837711.  Google Scholar

[31]

N. Komodakis, Image inpainting, IEEE Conference on Computer Vision and Pattern Recognition, 1 (2006), 442-452.   Google Scholar

[32]

R.-J. Lai and J. Li, Manifold based low-rank regularization for image restoration and semi-supervised learning, Journal of Scientific Computing, 74 (2018), 1241-1263.  doi: 10.1007/s10915-017-0492-x.  Google Scholar

[33]

J. I. Latorre, Image Compression and Entanglement, Computer Science, 2005. Google Scholar

[34]

F. Li, M. K. Ng and R. J. Plemmons, Coupled segmentation and denoising/deblurring models for hyperspectral material identification, Numerical Linear Algebra with Applications, 19 (2012), 153-173.  doi: 10.1002/nla.750.  Google Scholar

[35]

Y.-P. Liu, Z. Long and C. Zhu, Image completion using low tensor tree rank and total variation minimization, IEEE Transactions on Multimedia, 21 (2019), 338-350.  doi: 10.1109/TMM.2018.2859026.  Google Scholar

[36]

Y.-Y. Liu, F.-H. Shang, L.-C. Jiao, J. Cheng and H. Cheng, Trace norm regularized CANDECOMP/PARAFAC decomposition with missing data, IEEE Transactions on Cybernetics, 45 (2015), 2437-2448.  doi: 10.1109/TCYB.2014.2374695.  Google Scholar

[37]

Y.-P. Liu, Z. Long, H.-Y. Huang and C. Zhu, Low CP rank and tucker rank tensor completion for estimating missing components in image data, IEEE Transactions on Circuits and Systems for Video Technology, 30 (2020), 944-954.  doi: 10.1109/TCSVT.2019.2901311.  Google Scholar

[38]

J. Liu, P. Musialski, P. Wonka and J. Ye, Tensor completion for estimating missing values in visual data, IEEE Transactions on Pattern Analysis and Machine Intelligence, 35 (2013), 208-220.  doi: 10.1109/TPAMI.2012.39.  Google Scholar

[39]

C.-Y. Lu, J.-S. Feng, Y.-D. Chen, W. Liu, Z.-C. Lin and S.-C. Yan, Tensor robust principal component analysis: Exact recovery of corrupted low-rank tensors via convex optimization, IEEE Conference on Computer Vision and Pattern Recognition, (2016), 5249–5257. doi: 10.1109/CVPR.2016.567.  Google Scholar

[40]

C.-Y. Lu, J.-S. Feng, Z.-C. Lin and S.-C. Yan, Exact Low Tubal Rank Tensor Recovery from Gaussian Measurements, International Joint Conference on Artificial Intelligence, 2018. doi: 10.24963/ijcai.2018/347.  Google Scholar

[41]

I. V. Oseledets, Tensor-train decomposition, SIAM Journal on Scientific Computing, 33 (2011), 2295-2317.  doi: 10.1137/090752286.  Google Scholar

[42]

S. Osher, Z.-Q. Shi and W. Zhu, Low dimensional manifold model for image processing, SIAM Journal on Imaging Sciences, 10 (2017), 1669-1690.  doi: 10.1137/16M1058686.  Google Scholar

[43]

E. E. Papalexakis, C. Faloutsos and N. D. Sidiropoulos, Tensors for data mining and data fusion: Models, applications, and scalable algorithms, ACM Transactions on Intelligent Systems and Technology, 8 (2017), 16: 1–16: 44. doi: 10.1145/2915921.  Google Scholar

[44]

Z.-Q. Shi, S. Osher and W. Zhu, Weighted nonlocal laplacian on interpolation from sparse data, Journal of Scientific Computing, 73 (2017), 1164-1177.  doi: 10.1007/s10915-017-0421-z.  Google Scholar

[45]

N. D. Sidiropoulos, L. De Lathauwer, X. Fu, K. Huang, E. E. Papalexakis and C. Faloutsos, Tensor decomposition for signal processing and machine learning, IEEE Transactions on Signal Processing, 65 (2017), 3551-3582.  doi: 10.1109/TSP.2017.2690524.  Google Scholar

[46]

G. W. Stewart, Matrix Algorithms, Vol I: Basic Decomposition, SIAM, 2001. doi: 10.1137/1.9780898718058.  Google Scholar

[47]

W. Wang, V. Aggarwal and S. Aeron, Tensor completion by alternating minimization under the tensor train (TT) model, preprint, arXiv: 1609.05587. Google Scholar

[48]

Y. Wang, J.-J. Peng, Q. Zhao, Y. Leung, X.-L. Zhao and D.-Y. Meng, Hyperspectral image restoration via total variation regularized low-rank tensor decomposition, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 11 (2018), 1227-1243.   Google Scholar

[49]

Y. Wang, D.-Y. Meng and M. Yuan, Sparse recovery: From vectors to tensors, National Science Review, 5 (2018), 756-767.  doi: 10.1093/nsr/nwx069.  Google Scholar

[50]

T. Xie, S.-T. Li, L.-Y. Fang and L.-C. Liu, Tensor completion via nonlocal low-rank regularization, IEEE Transactions on Cybernetics, 49 (2019), 2344-2354.  doi: 10.1109/TCYB.2018.2825598.  Google Scholar

[51]

Z.-M. Xing, M.-Y. Zhou, A. Castrodad, G. Sapiro and L. Carin, Dictionary learning for noisy and incomplete hyperspectral images, SIAM Journal on Imaging Sciences, 5 (2012), 33-56.  doi: 10.1137/110837486.  Google Scholar

[52]

B. Xiong, Q.-G. Liu, J.-J. Xiong, S.-Q. Li, S.-S. Wang and D. Liang, Field-of-experts filters guided tensor completion, IEEE Transactions on Multimedia, 20 (2018), 2316-2329.  doi: 10.1109/TMM.2018.2806225.  Google Scholar

[53]

R.-T. Xu, Y. Xu and Y.-H. Quan, Factorized tensor dictionary learning for visual tensor data completion, IEEE Transactions on Multimedia, (2020), 1–14. Google Scholar

[54]

Y.-Y. Xu, R.-R. Hao, W.-T. Yin and Z.-X. Su, Parallel matrix factorization for low-rank tensor completion, Inverse Problems and Imaging, 9 (2017), 601-624.  doi: 10.3934/ipi.2015.9.601.  Google Scholar

[55]

J.-H. Yang, X.-L. Zhao, T.-H. Ma, M. Ding and T.-Z. Huang, Tensor train rank minimization with hybrid smoothness for visual data recovery, Applied Mathematical Modelling, 81 (2020), 711-726.  doi: 10.1016/j.apm.2020.01.039.  Google Scholar

[56]

J.-H. Yang, X.-L. Zhao, T.-H. Ma, Y. Chen, T.-Z. Huang and M. Ding, Remote sensing image destriping using unidirectional hybrid total variation and nonconvex low-rank regularization, Journal of Computational and Applied Mathematics, 363 (2020), 124-144.  doi: 10.1016/j.cam.2019.06.004.  Google Scholar

[57]

T. Yokota, Q.-B. Zhao, C. Li and A. Cichocki, Smooth PARAFAC decomposition for tensor completion, IEEE Transactions on Signal Processing, 64 (2016), 5423-5436.  doi: 10.1109/TSP.2016.2586759.  Google Scholar

[58]

Q.-B. Zhao, L-Q. Zhang and A. Cichocki, Bayesian CP factorization of incomplete tensors with automatic rank determination, IEEE Transactions on Pattern Analysis and Machine Intelligence, 37 (2015), 1751-1763.  doi: 10.1109/TPAMI.2015.2392756.  Google Scholar

[59]

X.-L. Zhao, W. Wang, T.-Y. Zeng, T.-Z. Huang and M. K. Ng, Total variation structured total least squares method for image restoration, SIAM Journal on Scientific Computing, 35 (2013), 1304-1320.  doi: 10.1137/130915406.  Google Scholar

[60]

X.-J. Zhang, A nonconvex relaxation approach to low-rank tensor completion, IEEE Transactions on Neural Networks and Learning Systems, 30 (2019), 1659-1671.  doi: 10.1109/TNNLS.2018.2872583.  Google Scholar

[61]

K.-B. Zhang, X.-B. Gao, D. -C.Tao and X.-L. Li, Single image super-resolution with non-local means and steering kernel regression, IEEE Transactions on Image Processing, 21 (2012), 4544-4556.  doi: 10.1109/TIP.2012.2208977.  Google Scholar

[62]

Z. Zhang, G. Ely and S. Aeron, Exact tensor completion using t-SVD, IEEE Transactions on Signal Processing, 65 (2017), 1511-1526.  doi: 10.1109/TSP.2016.2639466.  Google Scholar

[63]

Y.-B. Zheng, T.-Z. Huang, X.-L. Zhao, Y. Chen and W. He, Double-factor-regularized low-rank tensor factorization for mixed noise removal in hyperspectral image, IEEE Transactions on Geoscience and Remote Sensing, (2020), 1–15. Google Scholar

Figure 1.  Comparison of TT low-rankness of higher-order tensors generated by KA and NSS. (a-1, 2, 3) the augmented tensor, the original data, and an example of grouped tensors. (b-1) to (b-8) the distribution of singular values of the mode-1 to mode-8 canonical matricizations of the augmented tensor (a-1) and the average ratio of singular values larger than $ 1\% $ of the corresponding largest one is $ 25.2\% $. (c-1, 2, 3) the distribution of singular values of the mode-1, mode-2, and mode-3 canonical matricizations of the grouped tensor (a-3) and the average ratio of singular values larger than $ 1\% $ of the corresponding largest one is $ 7.1\% $. (d-1, 2, 3, 4) the observed data, the recovered results by SiLRTC-TT, TMac-TT, and the proposed method
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 2.  Flowchart of the proposed completion framework
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 3.  Original images
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 4.  The results of testing color images with $ SR = 0.2 $ recovered by different methods. The first three rows and second three rows represent random sampling and tube sampling, respectively. From left to right: (a) the observed image, the results by (b) HaLRTC, (c) tSVD, (d) SiLRTC-TT, (e) TMac-TT, (f) NL-TT, and (g) the original image
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 5.  The PSNR and SSIM values of the reconstructed color image results for random missing entries by different methods
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 6.  The results of testing color images with structural missing entries recovered by different methods. From left to right: (a) the observed image, the results by (b) HaLRTC, (c) tSVD, (d) SiLRTC-TT, (e) TMac-TT, (f) NL-TT, and (g) the original image
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 7.  The results of one band of testing MSIs with $ SR = 0.1 $ recovered by different methods. From left to right: (a) the observed image, the results by (b) HaLRTC, (c) tSVD, (d) SiLRTC-TT, (e) TMac-TT, (f) NL-TT, and (g) the original image
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 8.  The PSNR and SSIM values of all bands of the reconstructed MSIs with $ SR = 0.2 $ recovered by different methods
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 9.  Comparison of the PSNR values by different methods on the dataset CAVE with $ SR = 0.1 $
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 10.  The results of two frames of testing color videos recovered by different methods. The first (third) and second (fourth) rows: the results of color videos bus (mobile), respectively. From left to right: (a) the observed image, the results by (b) HaLRTC, (c) tSVD, (d) SiLRTC-TT, (e) TMac-TT, (f) NL-TT, and (g) the original image
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 11.  The PSNR and SSIM values of all frames of color videos recovered by different methods
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 12.  The PSNR and SSIM curves as the function of the number of the similar cube $ h $. (a) change in the PSNR value, (b) change in the SSIM value
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 13.  The PSNR and SSIM curves as the function of the number of the cube size $ s $. (a) change in the PSNR value, (b) change in the SSIM value
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 14.  The relative error and objective function values curve versus the iteration number for grouped tubes. (a) grouped tensors, (b) change in the relative error value, (c) change in the objective function value
Figure Options

Download full-size image

Download as PowerPoint slide

Figure 15.  The results of testing color images with random sampling ($ SR = 0.3 $) recovered by different methods. From left to right: (a) the observed image, the results by (b) LDMM, (c) WNLL, (d) MLR, (e) NL-TT, and (f) the original image
Figure Options

Download full-size image

Download as PowerPoint slide

Table 1.  The PSNR, SSIM values, and running time (in seconds) obtained by HaLRTC, tSVD, SiLRTC-TT, TMac-TT and NL-TT for color image data with different sampling rates (SRs). The first three rows and second three rows represent random sampling and tube sampling, respectively
Image SR 0.1 0.2 0.3 0.4
Method PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time
lena HaLRTC 19.29 0.4151 5.52 23.10 0.6047 4.65 25.68 0.7311 8.65 28.00 0.8205 7.83
tSVD 19.55 0.3500 87.65 23.33 0.5572 88.40 26.08 0.7033 88.80 28.60 0.8066 87.16
SiLRTC-TT 21.67 0.5954 39.17 24.80 0.7366 27.36 27.01 0.8226 20.32 28.90 0.8782 16.11
TMac-TT 24.25 0.6829 67.99 27.22 0.8097 65.13 28.87 0.8584 27.99 30.22 0.8902 20.38
NL-TT 26.52 0.8124 192.95 30.09 0.8970 121.74 32.02 0.9309 90.28 33.87 0.9528 76.41
airplane HaLRTC 19.80 0.4621 6.80 23.18 0.6437 4.69 25.62 0.7614 3.94 27.97 0.8399 7.36
tSVD 19.87 0.4196 84.57 23.30 0.6139 87.18 25.86 0.7387 87.63 28.25 0.8258 87.57
SiLRTC-TT 20.81 0.6072 32.20 23.42 0.7361 25.32 25.62 0.8213 20.04 27.55 0.8768 17.06
TMac-TT 22.46 0.6766 7.62 25.81 0.8105 60.23 27.67 0.8622 25.36 28.97 0.8915 16.93
NL-TT 24.33 0.7840 299.77 28.33 0.8929 109.30 30.29 0.9268 85.13 31.99 0.9489 107.19
monarch HaLRTC 17.12 0.4381 8.20 19.59 0.6069 4.25 21.89 0.7404 3.66 24.20 0.8271 3.33
tSVD 17.14 0.3372 91.07 19.98 0.5462 87.39 22.60 0.6980 88.03 25.23 0.8023 89.33
SiLRTC-TT 17.95 0.5784 38.01 20.32 0.7196 30.94 22.38 0.8100 26.44 24.39 0.8702 22.52
TMac-TT 19.21 0.6621 185.13 22.45 0.7912 104.59 24.86 0.8505 76.80 27.24 0.9046 81.50
NL-TT 22.22 0.8307 587.32 25.42 0.9140 379.66 27.95 0.9496 306.90 30.74 0.9729 246.85
lena HaLRTC 17.54 0.2942 6.37 20.97 0.4651 7.06 23.59 0.6144 5.38 25.88 0.7272 4.01
tSVD 17.88 0.2570 88.13 20.85 0.4186 99.70 23.29 0.5676 93.58 25.50 0.6857 90.83
SiLRTC-TT 20.90 0.5462 47.73 23.61 0.6830 33.84 25.69 0.7732 25.55 27.35 0.8353 19.25
TMac-TT 21.62 0.5629 15.30 24.60 0.7193 41.10 26.22 0.7764 17.78 27.55 0.8392 51.87
NL-TT 23.94 0.7351 223.34 27.45 0.8459 172.99 29.33 0.8928 93.35 31.38 0.9259 72.19
airplane HaLRTC 17.81 0.3050 5.93 20.77 0.4847 6.32 23.15 0.6214 4.71 25.29 0.7289 4.14
tSVD 17.97 0.2900 85.67 20.66 0.4588 85.71 22.97 0.5926 84.86 25.06 0.7029 89.59
SiLRTC-TT 20.20 0.5570 39.31 22.49 0.6809 28.91 24.33 0.7661 23.86 26.09 0.8298 20.87
TMac-TT 21.06 0.6169 35.50 23.15 0.7114 21.76 24.41 0.7729 19.68 26.17 0.8416 45.12
NL-TT 22.45 0.7255 175.78 25.25 0.8210 103.25 27.29 0.8749 81.90 29.24 0.9149 67.92
monarch HaLRTC 16.04 0.3424 6.13 18.28 0.5031 6.17 20.12 0.6363 4.62 21.93 0.7401 4.03
tSVD 16.33 0.2786 84.71 18.21 0.4312 86.92 19.90 0.5620 115.32 21.65 0.6791 88.34
SiLRTC-TT 17.46 0.5472 110.83 19.48 0.6695 34.32 21.19 0.7606 41.57 22.83 0.8290 25.77
TMac-TT 15.12 0.3466 142.62 18.66 0.6710 144.85 21.74 0.7739 73.79 23.49 0.8282 32.73
NL-TT 18.07 0.6564 147.54 22.33 0.8462 142.52 24.53 0.9086 108.21 26.25 0.9391 89.87
Table Options

Download as excel

Image SR 0.1 0.2 0.3 0.4
Method PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time
lena HaLRTC 19.29 0.4151 5.52 23.10 0.6047 4.65 25.68 0.7311 8.65 28.00 0.8205 7.83
tSVD 19.55 0.3500 87.65 23.33 0.5572 88.40 26.08 0.7033 88.80 28.60 0.8066 87.16
SiLRTC-TT 21.67 0.5954 39.17 24.80 0.7366 27.36 27.01 0.8226 20.32 28.90 0.8782 16.11
TMac-TT 24.25 0.6829 67.99 27.22 0.8097 65.13 28.87 0.8584 27.99 30.22 0.8902 20.38
NL-TT 26.52 0.8124 192.95 30.09 0.8970 121.74 32.02 0.9309 90.28 33.87 0.9528 76.41
airplane HaLRTC 19.80 0.4621 6.80 23.18 0.6437 4.69 25.62 0.7614 3.94 27.97 0.8399 7.36
tSVD 19.87 0.4196 84.57 23.30 0.6139 87.18 25.86 0.7387 87.63 28.25 0.8258 87.57
SiLRTC-TT 20.81 0.6072 32.20 23.42 0.7361 25.32 25.62 0.8213 20.04 27.55 0.8768 17.06
TMac-TT 22.46 0.6766 7.62 25.81 0.8105 60.23 27.67 0.8622 25.36 28.97 0.8915 16.93
NL-TT 24.33 0.7840 299.77 28.33 0.8929 109.30 30.29 0.9268 85.13 31.99 0.9489 107.19
monarch HaLRTC 17.12 0.4381 8.20 19.59 0.6069 4.25 21.89 0.7404 3.66 24.20 0.8271 3.33
tSVD 17.14 0.3372 91.07 19.98 0.5462 87.39 22.60 0.6980 88.03 25.23 0.8023 89.33
SiLRTC-TT 17.95 0.5784 38.01 20.32 0.7196 30.94 22.38 0.8100 26.44 24.39 0.8702 22.52
TMac-TT 19.21 0.6621 185.13 22.45 0.7912 104.59 24.86 0.8505 76.80 27.24 0.9046 81.50
NL-TT 22.22 0.8307 587.32 25.42 0.9140 379.66 27.95 0.9496 306.90 30.74 0.9729 246.85
lena HaLRTC 17.54 0.2942 6.37 20.97 0.4651 7.06 23.59 0.6144 5.38 25.88 0.7272 4.01
tSVD 17.88 0.2570 88.13 20.85 0.4186 99.70 23.29 0.5676 93.58 25.50 0.6857 90.83
SiLRTC-TT 20.90 0.5462 47.73 23.61 0.6830 33.84 25.69 0.7732 25.55 27.35 0.8353 19.25
TMac-TT 21.62 0.5629 15.30 24.60 0.7193 41.10 26.22 0.7764 17.78 27.55 0.8392 51.87
NL-TT 23.94 0.7351 223.34 27.45 0.8459 172.99 29.33 0.8928 93.35 31.38 0.9259 72.19
airplane HaLRTC 17.81 0.3050 5.93 20.77 0.4847 6.32 23.15 0.6214 4.71 25.29 0.7289 4.14
tSVD 17.97 0.2900 85.67 20.66 0.4588 85.71 22.97 0.5926 84.86 25.06 0.7029 89.59
SiLRTC-TT 20.20 0.5570 39.31 22.49 0.6809 28.91 24.33 0.7661 23.86 26.09 0.8298 20.87
TMac-TT 21.06 0.6169 35.50 23.15 0.7114 21.76 24.41 0.7729 19.68 26.17 0.8416 45.12
NL-TT 22.45 0.7255 175.78 25.25 0.8210 103.25 27.29 0.8749 81.90 29.24 0.9149 67.92
monarch HaLRTC 16.04 0.3424 6.13 18.28 0.5031 6.17 20.12 0.6363 4.62 21.93 0.7401 4.03
tSVD 16.33 0.2786 84.71 18.21 0.4312 86.92 19.90 0.5620 115.32 21.65 0.6791 88.34
SiLRTC-TT 17.46 0.5472 110.83 19.48 0.6695 34.32 21.19 0.7606 41.57 22.83 0.8290 25.77
TMac-TT 15.12 0.3466 142.62 18.66 0.6710 144.85 21.74 0.7739 73.79 23.49 0.8282 32.73
NL-TT 18.07 0.6564 147.54 22.33 0.8462 142.52 24.53 0.9086 108.21 26.25 0.9391 89.87
Table 2.  The PSNR, SSIM values, and running time (in seconds) obtained by HaLRTC, tSVD, SiLRTC-TT, TMac-TT and NL-TT for color image data with structural missing entries
Method HaLRTC tSVD SiLRTC-TT TMac-TT NL-TT
Image PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time
house 36.44 0.9707 6.98 36.14 0.9681 86.86 38.52 0.9793 14.05 38.03 0.9740 11.93 45.34 0.9906 26.46
facade 12.95 0.5681 0.12 12.95 0.5681 83.16 28.14 0.9062 28.01 27.50 0.8947 9.24 29.60 0.9357 416.18
sailboat 26.49 0.8700 4.63 26.69 0.8696 86.10 26.53 0.8838 39.29 26.40 0.8995 8.44 27.86 0.9370 171.55
barbara 32.44 0.9580 4.83 32.44 0.9579 86.28 33.99 0.9681 20.24 33.29 0.9654 6.87 37.56 0.9867 40.36
peppers 31.64 0.9595 2.22 31.53 0.9551 85.85 32.59 0.9676 24.55 32.77 0.9651 9.27 36.33 0.9862 46.86
Average 27.99 0.8653 3.76 27.95 0.8638 85.65 31.95 0.9410 25.23 31.60 0.9397 9.15 35.34 0.9672 140.28
Table Options

Download as excel

Method HaLRTC tSVD SiLRTC-TT TMac-TT NL-TT
Image PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time
house 36.44 0.9707 6.98 36.14 0.9681 86.86 38.52 0.9793 14.05 38.03 0.9740 11.93 45.34 0.9906 26.46
facade 12.95 0.5681 0.12 12.95 0.5681 83.16 28.14 0.9062 28.01 27.50 0.8947 9.24 29.60 0.9357 416.18
sailboat 26.49 0.8700 4.63 26.69 0.8696 86.10 26.53 0.8838 39.29 26.40 0.8995 8.44 27.86 0.9370 171.55
barbara 32.44 0.9580 4.83 32.44 0.9579 86.28 33.99 0.9681 20.24 33.29 0.9654 6.87 37.56 0.9867 40.36
peppers 31.64 0.9595 2.22 31.53 0.9551 85.85 32.59 0.9676 24.55 32.77 0.9651 9.27 36.33 0.9862 46.86
Average 27.99 0.8653 3.76 27.95 0.8638 85.65 31.95 0.9410 25.23 31.60 0.9397 9.15 35.34 0.9672 140.28
Table 3.  The average PSNR, SSIM values, and running time (in seconds) obtained by HaLRTC, tSVD, SiLRTC-TT, TMac-TT and NL-TT for MSIs with different SRs
Image SR 0.05 0.1 0.2
Method PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time
toy HaLRTC 20.14 0.6519 8.42 23.99 0.7790 6.69 28.91 0.8994 7.27
tSVD 25.89 0.7680 264.95 30.34 0.8844 271.40 36.57 0.9602 300.93
SiLRTC-TT 22.36 0.7138 379.68 25.81 0.8392 188.36 30.44 0.9433 513.80
TMac-TT 27.28 0.8329 232.52 32.37 0.9317 143.12 35.74 0.9669 50.91
NL-TT 29.58 0.9243 1414.34 34.44 0.9730 803.33 38.72 0.9899 557.72
feathers HaLRTC 20.66 0.6422 9.96 24.26 0.7720 14.14 28.81 0.8876 11.07
tSVD 25.15 0.6886 274.69 29.29 0.8266 340.60 34.82 0.9265 264.27
SiLRTC-TT 22.86 0.7196 247.87 26.32 0.8417 232.09 31.11 0.9411 374.05
TMac-TT 27.29 0.7611 58.00 32.12 0.9190 216.36 36.63 0.9631 62.90
NL-TT 29.61 0.9102 1080.65 34.76 0.9699 860.40 39.56 0.9879 607.91
superballs HaLRTC 23.28 0.7661 20.48 28.63 0.8621 9.71 34.10 0.9426 11.93
tSVD 28.24 0.7636 267.41 32.39 0.8663 270.40 38.20 0.9564 270.92
SiLRTC-TT 26.27 0.8290 289.90 29.79 0.9087 157.46 34.03 0.9651 379.94
TMac-TT 29.97 0.8343 60.71 33.90 0.9346 63.73 40.19 0.9803 109.14
NL-TT 32.93 0.9507 1150.35 37.25 0.9812 666.11 42.67 0.9939 409.49
Table Options

Download as excel

Image SR 0.05 0.1 0.2
Method PSNR SSIM Time PSNR SSIM Time PSNR SSIM Time
toy HaLRTC 20.14 0.6519 8.42 23.99 0.7790 6.69 28.91 0.8994 7.27
tSVD 25.89 0.7680 264.95 30.34 0.8844 271.40 36.57 0.9602 300.93
SiLRTC-TT 22.36 0.7138 379.68 25.81 0.8392 188.36 30.44 0.9433 513.80
TMac-TT 27.28 0.8329 232.52 32.37 0.9317 143.12 35.74 0.9669 50.91
NL-TT 29.58 0.9243 1414.34 34.44 0.9730 803.33 38.72 0.9899 557.72
feathers HaLRTC 20.66 0.6422 9.96 24.26 0.7720 14.14 28.81 0.8876 11.07
tSVD 25.15 0.6886 274.69 29.29 0.8266 340.60 34.82 0.9265 264.27
SiLRTC-TT 22.86 0.7196 247.87 26.32 0.8417 232.09 31.11 0.9411 374.05
TMac-TT 27.29 0.7611 58.00 32.12 0.9190 216.36 36.63 0.9631 62.90
NL-TT 29.61 0.9102 1080.65 34.76 0.9699 860.40 39.56 0.9879 607.91
superballs HaLRTC 23.28 0.7661 20.48 28.63 0.8621 9.71 34.10 0.9426 11.93
tSVD 28.24 0.7636 267.41 32.39 0.8663 270.40 38.20 0.9564 270.92
SiLRTC-TT 26.27 0.8290 289.90 29.79 0.9087 157.46 34.03 0.9651 379.94
TMac-TT 29.97 0.8343 60.71 33.90 0.9346 63.73 40.19 0.9803 109.14
NL-TT 32.93 0.9507 1150.35 37.25 0.9812 666.11 42.67 0.9939 409.49
Table 4.  The PSNR and SSIM values obtained by LDMM, WNLL, MLR, and NL-TT for color image data with different SRs
Image SR 0.1 0.2 0.3 0.4
Method PSNR SSIM PSNR SSIM PSNR SSIM PSNR SSIM
lena LDMM 22.20 0.6202 26.52 0.7944 27.74 0.8457 29.95 0.8879
WNLL 26.24 0.8043 28.08 0.8534 29.29 0.8855 30.35 0.9071
MLR 26.47 0.8052 29.03 0.8747 30.74 0.9119 32.14 0.9333
NL-TT 26.52 0.8124 30.09 0.8970 32.02 0.9309 33.87 0.9528
airplane LDMM 20.12 0.6230 24.03 0.7940 25.88 0.8532 28.59 0.8988
WNLL 23.75 0.7675 25.76 0.8264 27.00 0.8629 28.03 0.8891
MLR 24.14 0.7750 26.78 0.8595 28.62 0.9015 29.92 0.9260
NL-TT 24.33 0.7840 28.33 0.8929 30.29 0.9268 31.99 0.9489
monarch LDMM 18.61 0.6196 19.01 0.6463 22.34 0.8515 25.55 0.9176
WNLL 20.54 0.7584 22.61 0.8262 23.78 0.8619 24.93 0.8916
MLR 20.95 0.8030 23.73 0.8884 25.76 0.9291 27.48 0.9525
NL-TT 22.22 0.8307 25.42 0.9140 27.95 0.9496 30.74 0.9729
Table Options

Download as excel

Image SR 0.1 0.2 0.3 0.4
Method PSNR SSIM PSNR SSIM PSNR SSIM PSNR SSIM
lena LDMM 22.20 0.6202 26.52 0.7944 27.74 0.8457 29.95 0.8879
WNLL 26.24 0.8043 28.08 0.8534 29.29 0.8855 30.35 0.9071
MLR 26.47 0.8052 29.03 0.8747 30.74 0.9119 32.14 0.9333
NL-TT 26.52 0.8124 30.09 0.8970 32.02 0.9309 33.87 0.9528
airplane LDMM 20.12 0.6230 24.03 0.7940 25.88 0.8532 28.59 0.8988
WNLL 23.75 0.7675 25.76 0.8264 27.00 0.8629 28.03 0.8891
MLR 24.14 0.7750 26.78 0.8595 28.62 0.9015 29.92 0.9260
NL-TT 24.33 0.7840 28.33 0.8929 30.29 0.9268 31.99 0.9489
monarch LDMM 18.61 0.6196 19.01 0.6463 22.34 0.8515 25.55 0.9176
WNLL 20.54 0.7584 22.61 0.8262 23.78 0.8619 24.93 0.8916
MLR 20.95 0.8030 23.73 0.8884 25.76 0.9291 27.48 0.9525
NL-TT 22.22 0.8307 25.42 0.9140 27.95 0.9496 30.74 0.9729
[1]

Yangyang Xu, Ruru Hao, Wotao Yin, Zhixun Su. Parallel matrix factorization for low-rank tensor completion. Inverse Problems & Imaging, 2015, 9 (2) : 601-624. doi: 10.3934/ipi.2015.9.601
[2]

Changming Song, Yun Wang. Nonlocal latent low rank sparse representation for single image super resolution via self-similarity learning. Inverse Problems & Imaging, , () : -. doi: 10.3934/ipi.2021017
[3]

Wei Wan, Weihong Guo, Jun Liu, Haiyang Huang. Non-local blind hyperspectral image super-resolution via 4d sparse tensor factorization and low-rank. Inverse Problems & Imaging, 2020, 14 (2) : 339-361. doi: 10.3934/ipi.2020015
[4]

Tao Wu, Yu Lei, Jiao Shi, Maoguo Gong. An evolutionary multiobjective method for low-rank and sparse matrix decomposition. Big Data & Information Analytics, 2017, 2 (1) : 23-37. doi: 10.3934/bdia.2017006
[5]

Zhouchen Lin. A review on low-rank models in data analysis. Big Data & Information Analytics, 2016, 1 (2&3) : 139-161. doi: 10.3934/bdia.2016001
[6]

Russell E. Warren, Stanley J. Osher. Hyperspectral unmixing by the alternating direction method of multipliers. Inverse Problems & Imaging, 2015, 9 (3) : 917-933. doi: 10.3934/ipi.2015.9.917
[7]

Ke Wei, Jian-Feng Cai, Tony F. Chan, Shingyu Leung. Guarantees of riemannian optimization for low rank matrix completion. Inverse Problems & Imaging, 2020, 14 (2) : 233-265. doi: 10.3934/ipi.2020011
[8]

Simon Foucart, Richard G. Lynch. Recovering low-rank matrices from binary measurements. Inverse Problems & Imaging, 2019, 13 (4) : 703-720. doi: 10.3934/ipi.2019032
[9]

Sohana Jahan. Supervised distance preserving projection using alternating direction method of multipliers. Journal of Industrial & Management Optimization, 2020, 16 (4) : 1783-1799. doi: 10.3934/jimo.2019029
[10]

Jie-Wen He, Chi-Chon Lei, Chen-Yang Shi, Seak-Weng Vong. An inexact alternating direction method of multipliers for a kind of nonlinear complementarity problems. Numerical Algebra, Control & Optimization, 2021, 11 (3) : 353-362. doi: 10.3934/naco.2020030
[11]

Mingchao Zhao, You-Wei Wen, Michael Ng, Hongwei Li. A nonlocal low rank model for poisson noise removal. Inverse Problems & Imaging, 2021, 15 (3) : 519-537. doi: 10.3934/ipi.2021003
[12]

Dan Zhu, Rosemary A. Renaut, Hongwei Li, Tianyou Liu. Fast non-convex low-rank matrix decomposition for separation of potential field data using minimal memory. Inverse Problems & Imaging, 2021, 15 (1) : 159-183. doi: 10.3934/ipi.2020076
[13]

Yun Cai, Song Li. Convergence and stability of iteratively reweighted least squares for low-rank matrix recovery. Inverse Problems & Imaging, 2017, 11 (4) : 643-661. doi: 10.3934/ipi.2017030
[14]

Bingsheng He, Xiaoming Yuan. Linearized alternating direction method of multipliers with Gaussian back substitution for separable convex programming. Numerical Algebra, Control & Optimization, 2013, 3 (2) : 247-260. doi: 10.3934/naco.2013.3.247
[15]

Yuan Shen, Lei Ji. Partial convolution for total variation deblurring and denoising by new linearized alternating direction method of multipliers with extension step. Journal of Industrial & Management Optimization, 2019, 15 (1) : 159-175. doi: 10.3934/jimo.2018037
[16]

Yan Gu, Nobuo Yamashita. Alternating direction method of multipliers with variable metric indefinite proximal terms for convex optimization. Numerical Algebra, Control & Optimization, 2020, 10 (4) : 487-510. doi: 10.3934/naco.2020047
[17]

Zhongming Wu, Xingju Cai, Deren Han. Linearized block-wise alternating direction method of multipliers for multiple-block convex programming. Journal of Industrial & Management Optimization, 2018, 14 (3) : 833-855. doi: 10.3934/jimo.2017078
[18]

Rogelio Valdez. Self-similarity of the Mandelbrot set for real essentially bounded combinatorics. Discrete & Continuous Dynamical Systems, 2006, 16 (4) : 897-922. doi: 10.3934/dcds.2006.16.897
[19]

Yuan Shen, Xin Liu. An alternating minimization method for matrix completion problems. Discrete & Continuous Dynamical Systems - S, 2020, 13 (6) : 1757-1772. doi: 10.3934/dcdss.2020103
[20]

Qun Liu, Lihua Fu, Meng Zhang, Wanjuan Zhang. Two-dimensional seismic data reconstruction using patch tensor completion. Inverse Problems & Imaging, 2020, 14 (6) : 985-1000. doi: 10.3934/ipi.2020052

2019 Impact Factor: 1.373

Tools

Article outline

Figures and Tables

[Back to Top]

Copyright © 2021 American Institute of Mathematical Sciences