Sparsity-Constrained Deep Nonnegative Matrix Factorization for Hyperspectral Unmixing

Hao, Fang; Li, Aihua; Xu, Hongwei; Wang, Tao

doi:10.1109/lgrs.2018.2823425

Cited by 25 publications

(11 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where P refers to the proximal mapping. Secondly, we apply the forward-backward splitting framework to the DGsnMF problem, and ( 4) is rewritten as the minimization problem, min (9) where…”

Section: Forward-backward Splitting For Dgsnmfmentioning

confidence: 99%

“…Feng et al [10] present a sparsity-constrained deep nonnegative matrix factorization by enforcing L 1/2 constraint [18] and total variation for hyperspectral unmixing. It is worth noting that the deep matrix factorization methods obtain outstanding performance in many fields, including multi-view learning [38], remote sensing image processing [9], community detection [34], etc. However, the DMF with lots of variables is much more complex than the traditional NMF.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Converged Deep Graph Semi-NMF Algorithm for Learning Data Representation

Huang

Yang

et al. 2021

Circuits Syst Signal Process

View full text Add to dashboard Cite

Deep nonnegative matrix factorization (DMF) is a particularly useful technique for learning data representation in low-dimensional space. To further obtain the complex hidden information and keep the geometrical structures of the high-dimensional data, we propose a novel deep matrix factorization model with the graph regularization (called DGsnMF). For solving the model with multi-variables, we design a forward-backward splitting scheme. After that, the convergence analysis is attached to the proposed algorithm and it is proved to converge to a critical point. Empirical experiments on benchmark datasets show that the proposed method is superior to the compared methods.

show abstract

“…where P refers to the proximal mapping. Secondly, we apply the forward-backward splitting framework to the DGsnMF problem, and ( 4) is rewritten as the minimization problem, min (9) where…”

Section: Forward-backward Splitting For Dgsnmfmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Converged Deep Graph Semi-NMF Algorithm for Learning Data Representation

Huang

Yang

et al. 2021

Circuits Syst Signal Process

View full text Add to dashboard Cite

show abstract

“…The basic flowchart of the LBP-NAPCA pre-processing can be depicted as Figure 2. Plugging the pre-obtained candidate spatial information into the spatial regularization sparse unmixing model as an improved weight matrix, the propose algorithm can be constructed as (16). To solve the optimization problem (16) of the proposed method, alternating direction method of multipliers (ADMM) is used following the references [28,68], and we would revisit the whole optimization procedure in detail.…”

Section: Lbg-napca-based Sparse Unmixingmentioning

confidence: 99%

“…Traditionally, hyperspectral unmixing is divided into two stages, endmember selection and fractional abundance estimation [12]. In addition, blind source separation (BSS)-based unsupervised unmixing methods [13][14][15][16], such as independent component analysis [13] or non-negative matrix/tensor factorization based hyperspectral unmixing [14][15][16], also have been proven to be able to unmix highly mixed datasets and achieve comparable unmixing accuracy. However, due to the separation of the two stages in traditional spectral unmixing methods, the endmember estimation errors and fractional abundances estimation errors might be accumulated [17], which would lead to poor unmixing results; while the unsupervised unmixing techniques might also fail since they could extract virtual endmembers without any physical meaning, or these methods can only work under the hypothesis that there is pure pixel existing, which is hard to guarantee in reality [18][19][20][21].…”

Section: Introductionmentioning

confidence: 99%

Joint Local Block Grouping with Noise-Adjusted Principal Component Analysis for Hyperspectral Remote-Sensing Imagery Sparse Unmixing

2019

View full text Add to dashboard Cite

Spatial regularized sparse unmixing has been proved as an effective spectral unmixing technique, combining spatial information and standard spectral signatures known in advance into the traditional spectral unmixing model in the form of sparse regression. In a spatial regularized sparse unmixing model, spatial consideration acts as an important role and develops from local neighborhood pixels to global structures. However, incorporating spatial relationships will increase the computational complexity, and it is inevitable that some negative influences obtained by inaccurate estimated abundances’ spatial correlations will reduce the accuracy of the algorithms. To obtain a more reliable and efficient spatial regularized sparse unmixing results, a joint local block grouping with noise-adjusted principal component analysis for hyperspectral remote-sensing imagery sparse unmixing is proposed in this paper. In this work, local block grouping is first utilized to gather and classify abundant spatial information in local blocks, and noise-adjusted principal component analysis is used to compress these series of classified local blocks and select the most significant ones. Then the representative spatial correlations are drawn and replace the traditional spatial regularization in the spatial regularized sparse unmixing method. Compared with total variation-based and non-local means-based sparse unmixing algorithms, the proposed approach can yield comparable experimental results with three simulated hyperspectral data cubes and two real hyperspectral remote-sensing images.

show abstract

“…Since the traditional single-layer NMF may produce incorrect unmixing results, Feng et al [43] extended the single-layer NMF to deep NMF, which promoted the smooth segmentation of abundances. Besides, sparse constraints could be further enforced on each layer of the deep NMF [44]. The idea of subspace clustering has also been drawn into the NMF framework for unmixing [45], [46].…”

Section: Introductionmentioning

confidence: 99%