Novel Methods for Multilinear Data Completion and De-noising Based on Tensor-SVD
By Zemin Zhang, Gregory Ely, Shuchin Aeron, Ning Hao, Misha Elena Kilmer. 2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014
793 citations recorded by the source corpus.
In this paper we propose novel methods for completion (from limited samples) and de-noising of multilinear (tensor) data and as an application consider 3-D and 4- D (color) video data completion and de-noising. We exploit the recently proposed tensor-Singular Value Decomposition (t-SVD)[11]. Based on t-SVD, the notion of multilinear rank and a related tensor nuclear norm was proposed in [11] to characterize informational and structural complexity of multilinear data. We first show that videos with linear camera motion can be represented more efficiently using t-SVD compared to the approaches based on vectorizing or flattening of the tensors. Since efficiency in representation implies efficiency in recovery, we outline a tensor nuclear norm penalized algorithm for video completion from missing entries. Application of the proposed algorithm for video recovery from missing entries is shown to yield a superior performance over existing methods. We also consider the problem of tensor robust Principal Component Analysis (PCA) for de-noising 3-D video data from sparse random corruptions. We show superior performance of our method compared to the matrix robust PCA adapted to this setting as proposed in [4].
Subfields: Computational Mathematics, Computational Mechanics, Computer Vision and Pattern Recognition
View Paper: https://www.semanticscholar.org/paper/5c32e6850159ac722f7a63aece616b55a68bf289
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computational Mathematics
In Computational Mechanics
In Computer Vision and Pattern Recognition
Citations in the corpus
- 5D seismic data completion and denoising using a novel class of tensor decompositions (2015, methodology): “Recently, Zhang et al. (2014) demonstrate this information loss for the problem of 3D and 4D video data completion from missing pixels.”
- Kronecker-Basis-Representation Based Tensor Sparsity and Its Applications to Tensor Recovery (2018, methodology): “When N ¼ 3 , it’s also comparable to the t-SVD-based TC/TRPCA methods [42], [74], which need to compute I 3 (assuming I 3 (cid:10) N ¼ 3 ) SVDs for I 1 (cid:2) I 2 matrices each iteration.”
- Generalized tensor function via the tensor singular value decomposition based on the T-product (2019, background): “…33, 36, 45], low rank tensor recovery and robust tensor PCA [31, 33], data completion and denoising [11, 24, 25, 34, 36, 38, 39, 42, 47, 50, 52, 53, 54, 55], since the tensor T-product change problems to block circulant matrices which could be block diagonalizable by the fast Fourier…”
- Infrared Small Target Detection Based on Partial Sum of the Tensor Nuclear Norm (2019, methodology): “Derived from t-SVD, the tensor nuclear norm (TNN) was proposed in [51] and successfully applied to image recovery which had shown its advancement compared to SNN. Generally, minimizing the TNN may cause some unavoidable biases [46].”
- Self-supervised Knowledge Distillation Using Singular Value Decomposition (2018, methodology): “SVD is mainly used for dimension reduction or for extracting important information from feature maps [11,12,13].”
- On Unifying Multi-view Self-Representations for Clustering by Tensor Multi-rank Minimization (2016, methodology): “Now, we can define the tensor multi-rank as follows [16,17, 18] : Definition 6 (Tensor multi-rank) The multi-rank of X ∈ Rn1×n2×n3 is a vector r ∈ Rn3×1 with the i-th element equal to the rank of the i-th frontal slice of X f .”
- Tensor Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Tensors via Convex Optimization (2016, methodology): “The work [28] also uses ADMM to solve a similar problem as (6).”
- Tensor Robust Principal Component Analysis with a New Tensor Nuclear Norm (2018, background): “The above TNN definition is also different from existing works [19], [34], [27].”
- T-Jordan Canonical Form and T-Drazin Inverse Based on the T-Product (2019, background): “[56] The tensor T-product introduced by Kilmer [34] has been proved to be of great use in many areas, such as image processing [33, 34, 49, 55, 59, 70], computer vision [20, 64], signal processing [12, 41, 44, 58], low rank tensor recovery and robust tensor PCA [39, 41], data completion and denoisin”
- Tensor-Tensor Products with Invertible Linear Transforms (2015, methodology): “The t-product has been applied to image deblurring [1], face recognition [5], completion of seismic data [6], and tensor compression [7].”
- Essential Tensor Learning for Multi-View Spectral Clustering (2018, background): “Since t-SVD based tensor nuclear norm has been proven to be the tightest convex relaxation [17] to (cid:96) 1 -norm of the tensor multi-rank, so we adopt it. with the t-SVD based tensor nuclear norm, our model can well capture both the consistent and view-specific information among multiple views,…”
- Exact Tensor Completion Using t-SVD (2015, background): “L G ] 27 F e b 2015”