• P-ISSN 0974-6846 E-ISSN 0974-5645

Indian Journal of Science and Technology


Indian Journal of Science and Technology

Year: 2022, Volume: 15, Issue: 14, Pages: 608-618

Original Article

An Efficient Transform based Low Rank Tensor Completion to Extreme Visual Recovery

Received Date:31 January 2022, Accepted Date:20 February 2022, Published Date:09 April 2022


Objective: To propose an optimization approach in recovering of the corrupted tensors in the high dimensional real time data. Methods: The recovering of corrupted tensors is performed by low-rank tensor completion methods. The tensor decomposition methods are used in tensor completion methods. These Tensor decomposition methods; candecomp / parafac (CP), tucker and higher-order Singular Value Decomposition (HoSVD) are used to minimize the rank of a tensor data. The limitations are in finding the rank of a tensor. Findings: The recovered data using the lifting transform induced tensor- Singular Value Decomposition (t-SVD) technique were assessed utilizing the Peak Signal to Noise Ratio (PSNR), Structural Similarity (SSIM), Naturalness Image Quality Evaluator (NIQE), and Perceptual Image Quality Evaluator (PIQE). When compared to state-of-the-art approaches, the low rank assumption condition with the lifting transform consideration gave good data recovery for every missing ratio. Novelty: The missing data is calculated by lifting polyphase structures by utilizing the available data. The polyphase structures are splitting the value into equivalent multiple triangular matrices, these are processed with the t-SVD to have the better approximation tensor rank.

Keywords: Tensor Completion; Transformbased Optimization; 5/3 Lifting Wavelet Transform; Lowrank tensor completion; tSVD


  1. Xue S, Qiu W, Liu F, Jin X. Low-Rank Tensor Completion by Truncated Nuclear Norm Regularization. 2018 24th International Conference on Pattern Recognition (ICPR). 2018;p. 2600–2605. doi: 10.1109/ICPR.2018.8546008
  2. Su Y, Wu X, Liu W. Low-Rank Tensor Completion by Sum of Tensor Nuclear Norm Minimization. IEEE Access. 2019;7:134943–134953. Available from: https://dx.doi.org/10.1109/access.2019.2940664
  3. Zhang L, Song L, Du B, Zhang Y. Nonlocal Low-Rank Tensor Completion for Visual Data. IEEE Transactions on Cybernetics. 2021;51(2):673–685. Available from: https://dx.doi.org/10.1109/tcyb.2019.2910151
  4. Huang H, Liu Y, Liu J, Zhu C. Provable tensor ring completion. Signal Processing. 2020;171:107486. Available from: https://dx.doi.org/10.1016/j.sigpro.2020.107486
  5. Chen L, Jiang X, Liu X, Zhou Z. Logarithmic Norm Regularized Low-Rank Factorization for Matrix and Tensor Completion. IEEE Transactions on Image Processing. 2021;30:3434–3449. Available from: https://dx.doi.org/10.1109/tip.2021.3061908
  6. Liu J, Musialski P, Wonka P, Ye J. Tensor Completion for Estimating Missing Values in Visual Data. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2013;35(1):208–220. Available from: https://dx.doi.org/10.1109/tpami.2012.39
  7. Recht B, Fazel M, Parrilo PA. Guaranteed Minimum-Rank Solutions of Linear Matrix Equations via Nuclear Norm Minimization. SIAM Review. 2010;52(3):471–501. Available from: https://dx.doi.org/10.1137/070697835
  8. Hu Y, Zhang D, Ye J, Li X, He X. Fast and Accurate Matrix Completion via Truncated Nuclear Norm Regularization. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2013;35(9):2117–2130. Available from: https://dx.doi.org/10.1109/tpami.2012.271
  9. Dong J, Xue Z, Guan J, Han ZF, Wang W. Low rank matrix completion using truncated nuclear norm and sparse regularizer. Signal Processing: Image Communication. 2018;68:76–87. Available from: https://dx.doi.org/10.1016/j.image.2018.06.007
  10. Kilmer ME, Martin CD. Factorization strategies for third-order tensors. Linear Algebra and its Applications. 2011;435(3):641–658. Available from: https://dx.doi.org/10.1016/j.laa.2010.09.020
  11. Han ZF, Leung CS, Huang LT, So HC. Sparse and Truncated Nuclear Norm Based Tensor Completion. Neural Processing Letters. 2017;45(3):729–743. Available from: https://dx.doi.org/10.1007/s11063-016-9503-4
  12. Song G, Ng MK, Zhang X. Robust tensor completion using transformed tensor singular value decomposition. Numerical Linear Algebra with Applications. 2020;27(3). Available from: https://dx.doi.org/10.1002/nla.2299
  13. Ganesan T, Rajarajeswari P, Nayak SR, Bhatia AS. A novel genetic algorithm with CDF5/3 filter-based lifting scheme for optimal sensor placement. International Journal of Innovative Computing and Applications. 2021;12(2/3):67. doi: 10.1504/IJICA.2021.113746
  14. Wang JL, Huang TZ, Zhao XL, Jiang TX, Ng MK. Multi-Dimensional Visual Data Completion via Low-Rank Tensor Representation Under Coupled Transform. IEEE Transactions on Image Processing. 2021;30:3581–3596. Available from: https://dx.doi.org/10.1109/tip.2021.3062995
  15. Ding M, Huang TZ, Zhao XL, Ng MK, Ma TH. Tensor train rank minimization with nonlocal self-similarity for tensor completion. Inverse Problems & Imaging. 2021;15(3):475. Available from: https://dx.doi.org/10.3934/ipi.2021001
  16. Mittal A, Soundararajan R, Bovik AC. Making a “Completely Blind” Image Quality Analyzer. IEEE Signal Processing Letters. 2013;20(3):209–212. Available from: https://dx.doi.org/10.1109/lsp.2012.2227726
  17. Venkatanath N, Praneeth D, Chandrasekhar BM, Channappayya SS, Medasani SS. Blind image quality evaluation using perception based features. 2015 Twenty First National Conference on Communications (NCC). 2015;p. 1–6. doi: 10.1109/NCC.2015.7084843


© 2022 Jyothula & Talari.  This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Published By Indian Society for Education and Environment (iSee)


Subscribe now for latest articles and news.