Qualitative HD Image and Video Recovery via High-Order Tensor Augmentation and Completion

Hoang, Pham Minh; Tuan, Hoang Duong; Son, Tran Thai; Poor, H. Vincent

doi:10.1109/jstsp.2020.3042063

Cited by 4 publications

(11 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To circumvent the issue of unbalanced matrices, which makes the matrix-rank optimization based completion and compressed sensing deficient, as a remedy, it was proposed in [9] and [25] to represent these matrices by high-order and low-dimensional tensors for tensor completion. For tensor completion, the TT-based tensor decomposition [20], [21] has been used for avoiding the creation of only unbalanced matrix factors by Tucker decomposition, also known as higherorder singular value decomposition (HOSVD) [30].…”

Section: Completionmentioning

confidence: 99%

“…Naturally, the efficiency of this approach is heavily dependent on the capability of TA to capture all the correlations and entanglements among the matrix entries. We thus opt here for the most efficient known TA of [25], which works for matrices of flexible sizes, and it is capable of capturing the distinct correlations among coarse-grained textures. [25], the matrix X may be represented by an N -order tensor…”

Section: Completionmentioning

confidence: 99%

“…We thus opt here for the most efficient known TA of [25], which works for matrices of flexible sizes, and it is capable of capturing the distinct correlations among coarse-grained textures. [25], the matrix X may be represented by an N -order tensor…”

Section: Completionmentioning

confidence: 99%

“…More particularly, the quantum-state based ketaugmentation (KA) used in [24] has been shown to be very effective for tensor augmentation (TA) of representing 2 N ×2 N color images (3rd-order 2 N × 2 N × 3 tensors). By viewing TA as a problem of encoding two-dimensional (2D) indices by Ndigit words, a new TA was proposed in [25] for representing low-order tensors of flexible sizes and structures by high-order tensors, with the latter providing a completely new coarsegrained hierarchy of the former's entries.…”

Section: Introductionmentioning

confidence: 99%

“…• We lay the foundation for completing unbalanced and sparse low-order tensors, which is based on the TA concept [25] of representing low-order large-dimensional tensors by high-order low-dimensional tensors and then TT-decomposition for high-order tensors; • Based on our new results on the q -metric, we develop new high-performance algorithms for the completion of both unbalanced and sparse low-order tensors; • We develop new computationally efficient algorithms for the completion of unbalanced low-order tensors, which do not require singular-value-decomposition (SVD). The contributions of this work relative to previous related literature are shown in Table I.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Learning Unbalanced and Sparse Low-Order Tensors

Hoang

Tuan

Son

et al. 2022

IEEE Trans. Signal Process.

Self Cite

View full text Add to dashboard Cite

Efficient techniques are developed for completing unbalanced and sparse low-order tensors, which cannot be effectively completed by popular matrix-rank optimization based techniques such as compressed sensing and/or the q -matrixmetric. We use our previously developed 2D-index encoding technique for tensor augmentation in order to represent these incomplete low-order tensors by high-order but low-dimensional tensors with their modes building up a coarse-grained hierachy of correlations among the incomplete tensor entries. The concept of tensor-trains is then exploited for decomposing these augmented tensors into trains of balanced and sparse matrices for efficient completion. More explicitly, we develop powerful algorithms exhibiting an excellent performance vs. complexity trade-off, which are supported by numerical examples by relying on matrix data and third-order tensor data constituted by color images.Index Terms-Matrix and/or low-order tensor completion, tensor train decomposition, tensor train rank, q

show abstract

Section: Completionmentioning

confidence: 99%

Section: Completionmentioning

confidence: 99%

Section: Completionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Learning Unbalanced and Sparse Low-Order Tensors

Hoang

Tuan

Son

et al. 2022

IEEE Trans. Signal Process.

Self Cite

View full text Add to dashboard Cite

show abstract

TT-ViT: Vision Transformer Compression Using Tensor-Train Decomposition

Minh

Xuan

Thai

2022

Computational Collective Intelligence

View full text Add to dashboard Cite

HFRAS: design of a high-density feature representation model for effective augmentation of satellite images

Saini,

Garg,

Malik

et al. 2023

SIViP

View full text Add to dashboard Cite

Efficiently extracting features from satellite images is crucial for classification and post-processing activities. Many feature representation models have been created for this purpose. However, most of them either increase computational complexity or decrease classification efficiency. The proposed model in this paper initially collects a set of available satellite images and represents them via a hybrid of long short-term memory (LSTM) and gated recurrent unit (GRU) features. These features are processed via an iterative genetic algorithm, identifying optimal augmentation methods for the extracted feature sets. To analyse the efficiency of this optimization process, we model an iterative fitness function that assists in incrementally improving the classification process. The fitness function uses an accuracy & precision-based feedback mechanism, which helps in tuning the hyperparameters of the proposed LSTM & GRU feature extraction process. The suggested model used 100 k images, 60% allocated for training and 20% each designated for validation and testing purposes. The proposed model can increase classification precision by 16.1% and accuracy by 17.1% compared to conventional augmentation strategies. The model also showcased incremental accuracy enhancements for an increasing number of training image sets.

show abstract

Qualitative HD Image and Video Recovery via High-Order Tensor Augmentation and Completion

Cited by 4 publications

References 28 publications

Learning Unbalanced and Sparse Low-Order Tensors

Learning Unbalanced and Sparse Low-Order Tensors

TT-ViT: Vision Transformer Compression Using Tensor-Train Decomposition

HFRAS: design of a high-density feature representation model for effective augmentation of satellite images

Contact Info

Product

Resources

About