Denoising of Hyperspectral Images Employing Two-Phase Matrix Decomposition

Li, Qian; Li, Houqiang; Lu, Zhenbo; Lu, Qingbo; Li, Weiping

doi:10.1109/jstars.2014.2360409

Cited by 29 publications

(10 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1. Unlike the two-phase method proposed in [22], we combine the low-rank constraint and the TV regularization together in a unified mathematical framework and simultaneously detect the sparse noise and restore the HSI.…”

Section: Lrtv For Hsi Restorationmentioning

confidence: 99%

“…0196 In these approaches, in the first phase, a median-type filter is used to identify the impulse-corrupted (sparse noise) pixel set, and in the second phase, the data-fidelity term of the model utilizes only the uncorrupted pixels to restore the corrupted image. A two-phase approach has also been used in HSI mixed-noise removal [22] and achieved comparable results. In recent years, low-rank matrix factorization has been widely utilized as another powerful tool for image analysis, web search, and computer vision [23]- [25].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Total-Variation-Regularized Low-Rank Matrix Factorization for Hyperspectral Image Restoration

Wei

Zhang

et al. 2016

IEEE Trans. Geosci. Remote Sensing

555

301

View full text Add to dashboard Cite

In this paper, we present a spatial spectral hyperspectral image (HSI) mixed-noise removal method named total variation (TV)-regularized low-rank matrix factorization (LRTV). In general, HSIs are not only assumed to lie in a low-rank subspace from the spectral perspective but also assumed to be piecewise smooth in the spatial dimension. The proposed method integrates the nuclear norm, TV regularization, and L 1 -norm together in a unified framework. The nuclear norm is used to exploit the spectral low-rank property, and the TV regularization is adopted to explore the spatial piecewise smooth structure of the HSI. At the same time, the sparse noise, which includes stripes, impulse noise, and dead pixels, is detected by the L 1 -norm regularization. To tradeoff the nuclear norm and TV regularization and to further remove the Gaussian noise of the HSI, we also restrict the rank of the clean image to be no larger than the number of endmembers. A number of experiments were conducted in both simulated and real data conditions to illustrate the performance of the proposed LRTV method for HSI restoration.Index Terms-Hyperspectral image (HSI), low-rank matrix factorization, rank constraint, restoration, total variation (TV).

show abstract

Section: Lrtv For Hsi Restorationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Total-Variation-Regularized Low-Rank Matrix Factorization for Hyperspectral Image Restoration

Wei

Zhang

et al. 2016

IEEE Trans. Geosci. Remote Sensing

555

301

View full text Add to dashboard Cite

show abstract

“…The first outliers denoising step is however complex to tackle. Recently, and to the best of our knowledge, the most powerful techniques are based on the PCP algo-70 rithm (Principle Component Pursuit -[10]), which have been proposed in [6,11]. However, it is essential for the PCP algorithm to work that the contribution of the sources AS has low rank.…”

Section: Robust Bss Methods In the Literaturementioning

confidence: 99%

“…Important examples are: i) stripping noise or impulse noise in hyperspectral data [6]), ii) cosmic ray contamination in astronomical images [7], iii) point sources emissions in astrophysical data [8] to only name a few. Beyond instrumental or physical artifacts, it has been recently advocated that sparse deviations from the linear mixture model can be approximated with outliers mod-45 els in hyperspectral data [2].…”

mentioning

confidence: 99%

Blind separation of sparse sources in the presence of outliers

Chenot

Bobin

2017

Signal Processing

View full text Add to dashboard Cite

To cite this version:Cécile Chenot, Jérôme Bobin. Blind separation of sparse sources in the presence of outliers. Signal Processing, Elsevier, 2017, 138, pp.233 -243 robustness of this new algorithm with respect to aberrant outliers on a wide range of blind separation instances. In contrast to current robust BSS methods, the rAMCA algorithm is shown to perform very well when the number of observations is close or equal to the number of sources.

show abstract

“…To take advantage of the interband correlations in hyperspectral data, Chen et al [4] proposed a preprocessing method that makes use of the spatial piecewise-continuous nature of HSI to improve the accuracy of supervised classification techniques. Li et al [5] proposed an efficient noise reduction method for HSI based on matrix decomposition, which imposes no strict restriction on the pattern of the noise, nor does it require idealized assumptions on HSIs. Wang et al [6] proposed a HSI denoising algorithm based on oblique subspace, which considers the correlation between the signal and noise.…”

Section: Introductionmentioning

confidence: 99%

Gradient and Laplacian-Based Hyperspectral Anisotropic Diffusion

Wang

Yang

et al. 2015

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

View full text Add to dashboard Cite

To improve accuracy and efficiency of object detection and classification with hyperspectral imagery (HSI), we propose a novel smoothing algorithm by coupling of a Laplacianbased reaction term to a classical divergence-based anisotropic diffusion partial differential equation (PDE). In addition, an adaptive parameter is introduced to regularize this nonlinear reactiondiffusion PDE by explicitly integrating the interband correlations with the noise level of each band in HSI. It is also well-known that the interband correlations can be implicitly embedded into the diffusion coefficient of the divergence-based PDE, to allow a selective smoothing that reduces the local homogeneous area variability while preventing smoothing across edges. Therefore, the interband correlations in HSI are exploited in the proposed method in both explicit and implicit ways. As a result, our algorithm is more effective at controlling the behavior of the diffusion evolution when compared to previous multi/hyperspectral diffusion algorithms. The simulations based on both synthetic data and real hyperspectral remote sensing data show that our algorithm can improve the hyperspectral data quality in terms of both visual inspection and image quality indices.Index Terms-Anisotropic diffusion, hyperspectral, image smoothing, interband correlations, partial differential equation (PDE), remote sensing.

show abstract

Denoising of Hyperspectral Images Employing Two-Phase Matrix Decomposition

Cited by 29 publications

References 34 publications

Total-Variation-Regularized Low-Rank Matrix Factorization for Hyperspectral Image Restoration

Total-Variation-Regularized Low-Rank Matrix Factorization for Hyperspectral Image Restoration

Blind separation of sparse sources in the presence of outliers

Gradient and Laplacian-Based Hyperspectral Anisotropic Diffusion

Contact Info

Product

Resources

About