Existing methods for tensor completion (TC) have limited ability characterizing low-rank (LR) structures. To depict the complex hierarchical knowledge with implicit sparsity attributes hidden in a tensor, we propose new multilayer sparsity-based decomposition (MLSTD) (LRTC). The method encodes structured of by multiple-layer representation. Specifically, use CANDECOMP/PARAFAC (CP) model to decompose into an ensemble sum rank-1 tensors, and number components is easily interpreted as...

Sparse subspace clustering (SSC) has emerged as an effective approach for the automatic analysis of hyperspectral images (HSI). Traditional SSC-based approaches employ input HSI data a dictionary atoms, in terms which all samples are linearly represented. This leads to highly redundant dictionaries huge size, and computational complexity resulting optimization problems becomes prohibitive large-scale data. In this article, we propose scalable method, integrates learning concise robust...

Subspace clustering has achieved remarkable performance for hyperspectral image (HSI). However, existing methods are often computationally expensive and have limited ability to capture the intrinsic structural information of HSI. In this paper, we propose a prior-guided subspace method, which simultaneously incorporates local non-local spatial cluster prior information. Accordingly, three efficient regularizations developed. Considering connectivity pixels, an ℓ <sub...

Band selection (BS) reduces effectively the spectral dimension of a hyperspectral image (HSI) by selecting relatively few representative bands, which allows efficient processing in subsequent tasks. Existing unsupervised BS methods based on subspace clustering are built matrix-based models, where each band is reshaped as vector. They encode correlation data only mode (dimension) and neglect strong correlations between different modes, i.e., spatial modes mode. Another issue that...

Sparse subspace clustering (SSC), as an effective technique, has been widely applied in the remote sensing community, demonstrating a superior performance over traditional methods such k-means. In this paper, we propose unified framework for hyperspectral image (HSI) clustering, which incorporates spatial information and label SSC model, aiming at generating more precise similarity matrix. The is included through joint sparsity constraint on coefficient matrix of each local region. Pixels...

Band selection, which removes irrelevant bands from hyperspectral images (HSIs) and keeps essential spectral information contained in a relatively few bands, allows huge savings data storage, computation time, imaging hardware. In this article, we propose novel structural subspace clustering (STSC) method for band leverages the self-representation property of prior to learn cluster structure bands. Particularly, general model where coarse coefficients matrix derived is decomposed as...

Hearing impaired people have their own hearing loss characteristics and listening preferences. Therefore aid system should become more natural, humanized personalized, which requires the filterbank in aids provides flexible sound wave decomposition schemes, so that patients are likely to use most suitable scheme for compensation strategy. In this paper, a reconfigurable is proposed. The prototype filter first cosine modulated generate uniform subbands. Then by non-linear transformation...

Clustering algorithms play an essential and unique role in classification tasks, especially when annotated data are unavailable or very scarce. Current clustering approaches remote sensing mostly designed for a single source, such as hyperspectral image (HSI), while, nowadays, multisensor being routinely acquired. In this article, we propose multiview subspace model that exploits effectively the rich information from multiple features extracted either source (HSI) sources call generically...

Sparse representation has been extensively investigated for hyperspectral image (HSI) classification and led to substantial improvements in the performance over traditional methods, such as support vector machine (SVM). However, existing sparsity-based methods typically assume Gaussian noise, neglecting fact that HSIs are often corrupted by different types of noise practice. In this paper, we develop a robust model admits realistic mixed which includes sparse noise. We combine with prior on...

The multistage frequency-response masking (FRM) technique is widely used to reduce the complexity of a filter when transition bandwidth extremely small. In this brief, real generalized two-stage FRM without any constraint on subfilters or interpolation factors was proposed. New principles and equations were deduced determine design parameters. then jointly optimized using nonlinear optimization. Experiential results show that proposed algorithm obtains different solutions with conventional...

Sparse subspace clustering (SSC) has been widely applied in remote sensing demonstrating excellent performance. Recent extensions incorporate spatial information, typically via smoothness-enforcing regularization. We propose an alternative approach: a joint sparsity SSC model, where pixels within local region are enforced to select common set of samples the subspace-sparse representation. The corresponding optimization problem is solved by alternating direction method multipliers (ADMM)....

Supervised hyperspectral image (HSI) classification relies on accurate label information. However, it is not always possible to collect perfectly labels for training samples. This motivates the development of classifiers that are sufficiently robust some reasonable amounts errors in data labels. Despite growing importance this aspect, has been studied literature yet. In paper, we analyze effect erroneous sample probability distributions principal components HSIs, and provide way a...

Sparse subspace clustering (SSC) techniques provide the state-of-the-art in of hyperspectral images (HSIs). However, their computational complexity hinders applicability to large-scale HSIs. In this paper, we propose a SSC-based method, which can effectively process large HSIs while also achieving improved accuracy compared current SSC methods. We build our approach based on an emerging concept sketched clustering, was knowledge not explored at all imaging yet. Moreover, there are only...

Sparse representation based methods have demonstrated their superior performance in target detection tasks compared to more traditional approaches such as matched subspace detectors and adaptive detectors. However, the existing sparsity-based were mostly formulated for validated on a single imaging modality (sometimes with multiple spectral bands). In many application domains, including art investigation, multimodal data, acquired by different sensors are readily available, yet, efficient...

In this article, we propose a novel bilayer low-rankness measure and two models based on it to recover low-rank (LR) tensor. The global low rankness of underlying tensor is first encoded by LR matrix factorizations (MFs) the all-mode matricizations, which can exploit multiorientational spectral rankness. Presumably, factor matrices decomposition are LR, since local property exists in within-mode correlation. decomposed subspace, describe refined structures factor/subspace, new insight...

Hyperspectral images (HSIs), captured by different Earth observation airborne and space-borne systems, provide rich spectral information in hundreds of bands, enabling far better discrimination between ground materials that are often indistinguishable visible multi-spectral images. Clustering HSIs, which aims to unveil class patterns an unsupervised way, is highly important the interpretation HSI, especially when labelled data not available. A number HSI clustering methods have been...

Hyperspectral image (HSI) processing tasks frequently rely on Spatial-Spectral Total Variation (SSTV) to quantify the local smoothness of structures. However, conventional SSTV only considers a sparse structure gradient maps computed along spatial and spectral dimensions, while neglecting other correlations. To address this limitation, we introduce low-rank guided (LRSTV), which characterizes sparsity priors map simultaneously. Firstly, verify through numerical tests theoretical analysis...

We explore the potential of deep learning in digital painting analysis to facilitate condition reporting and support restoration treatments. address problem paint loss detection develop a multiscale system with dilated convolutions that enables large receptive field limited training parameters avoid overtraining. Our model handles efficiently multimodal data are typically acquired art investigation. As case study we use Ghent Altarpiece. results indicate huge proposed approach terms accuracy...

In the restoration process of classical paintings, one tasks is to map paint loss for documentation and analysing purposes. Because this such a sizable tedious job automatic techniques are highly on demand. The currently available tools allow only rough mapping areas while still requiring considerable manual work. We develop here learning method detection that makes use multimodal image acquisitions we apply it within current Ghent Altarpiece. Our neural network architecture inspired by...