Academic literature on the topic 'Hyperspectral super-Resolution'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the lists of relevant articles, books, theses, conference reports, and other scholarly sources on the topic 'Hyperspectral super-Resolution.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Journal articles on the topic "Hyperspectral super-Resolution":
1
Sun, Shasha, Wenxing Bao, Kewen Qu, Wei Feng, Xiaowu Zhang, and Xuan Ma. "Hyperspectral Image Super-Resolution Algorithm Based on Graph Regular Tensor Ring Decomposition." Remote Sensing 15, no. 20 (October 16, 2023): 4983. http://dx.doi.org/10.3390/rs15204983.
Abstract:
This paper introduces a novel hyperspectral image super-resolution algorithm based on graph-regularized tensor ring decomposition aimed at resolving the challenges of hyperspectral image super-resolution. This algorithm seamlessly integrates graph regularization and tensor ring decomposition, presenting an innovative fusion model that effectively leverages the spatial structure and spectral information inherent in hyperspectral images. At the core of the algorithm lies an iterative optimization process embedded within the objective function. This iterative process incrementally refines latent feature representations. It incorporates spatial smoothness constraints and graph regularization terms to enhance the quality of super-resolution reconstruction and preserve image features. Specifically, low-resolution hyperspectral images (HSIs) and high-resolution multispectral images (MSIs) are obtained through spatial and spectral downsampling, which are then treated as nodes in a constructed graph, efficiently fusing spatial and spectral information. By utilizing tensor ring decomposition, HSIs and MSIs undergo feature decomposition, and the objective function is formulated to merge reconstructed results with the original images. Through a multi-stage iterative optimization procedure, the algorithm progressively enhances latent feature representations, leading to super-resolution hyperspectral image reconstruction. The algorithm’s significant achievements are demonstrated through experiments, producing sharper, more detailed high-resolution hyperspectral images (HRIs) with an improved reconstruction quality and retained spectral information. By combining the advantages of graph regularization and tensor ring decomposition, the proposed algorithm showcases substantial potential and feasibility within the domain of hyperspectral image super-resolution.
2
Zhang, Jing, Renjie Zheng, Zekang Wan, Ruijing Geng, Yi Wang, Yu Yang, Xuepeng Zhang, and Yunsong Li. "Hyperspectral Image Super-Resolution Based on Feature Diversity Extraction." Remote Sensing 16, no. 3 (January 23, 2024): 436. http://dx.doi.org/10.3390/rs16030436.
Abstract:
Deep learning is an important research topic in the field of image super-resolution. Problematically, the performance of existing hyperspectral image super-resolution networks is limited by feature learning for hyperspectral images. Nevertheless, the current algorithms exhibit some limitations in extracting diverse features. In this paper, we address limitations to existing hyperspectral image super-resolution networks, focusing on feature learning challenges. We introduce the Channel-Attention-Based Spatial–Spectral Feature Extraction network (CSSFENet) to enhance hyperspectral image feature diversity and optimize network loss functions. Our contributions include: (a) a convolutional neural network super-resolution algorithm incorporating diverse feature extraction to enhance the network’s diversity feature learning by elevating the matrix rank, (b) a three-dimensional (3D) feature extraction convolution module, the Channel-Attention-Based Spatial–Spectral Feature Extraction Module (CSSFEM), to boost the network’s performance in both the spatial and spectral domains, (c) a feature diversity loss function designed based on the image matrix’s singular value to maximize element independence, and (d) a spatial–spectral gradient loss function introduced based on space and spectrum gradient values to enhance the reconstructed image’s spatial–spectral smoothness. In contrast to existing hyperspectral super-resolution algorithms, we used four evaluation indexes, PSNR, mPSNR, SSIM, and SAM, and our method showed superiority during testing with three common hyperspectral datasets.
3
Zhang, Yan, Lifu Zhang, Ruoxi Song, and Qingxi Tong. "A General Deep Learning Point–Surface Fusion Framework for RGB Image Super-Resolution." Remote Sensing 16, no. 1 (December 28, 2023): 139. http://dx.doi.org/10.3390/rs16010139.
Abstract:
Hyperspectral images are usually acquired in a scanning-based way, which can cause inconvenience in some situations. In these cases, RGB image spectral super-resolution technology emerges as an alternative. However, current mainstream spectral super-resolution methods aim to generate continuous spectral information at a very narrow range, limited to the visible light range. Some researchers introduce hyperspectral images as auxiliary data. But it is usually required that the auxiliary hyperspectral images have the same spatial range as RGB images. To address this issue, a general point–surface data fusion method is designed to achieve the RGB image spectral super-resolution goal in this paper, named GRSS-Net. The proposed method utilizes hyperspectral point data as auxiliary data to provide spectral reference information. Thus, the spectral super-resolution can extend the spectral reconstruction range according to spectral data. The proposed method utilizes compressed sensing theory as a fundamental physical mechanism and then unfolds the traditional hyperspectral image reconstruction optimization problem into a deep network. Finally, a high-spatial-resolution hyperspectral image can be obtained. Thus, the proposed method combines the non-linear feature extraction ability of deep learning and the interpretability of traditional physical models simultaneously. A series of experiments demonstrates that the proposed method can effectively reconstruct spectral information in RGB images. Meanwhile, the proposed method provides a framework of spectral super-resolution for different applications.
4
Chang, Pai-Chuan, Jhao-Ting Lin, Chia-Hsiang Lin, Po-Wei Tang, and Yangrui Liu. "Optimization-Based Hyperspectral Spatiotemporal Super-Resolution." IEEE Access 10 (2022): 37477–94. http://dx.doi.org/10.1109/access.2022.3163266.
5
Mianji, Fereidoun A., Ye Zhang, Humayun Karim Sulehria, Asad Babakhani, and Mohammad Reza Kardan. "Super-Resolution Challenges in Hyperspectral Imagery." Information Technology Journal 7, no. 7 (September 15, 2008): 1030–36. http://dx.doi.org/10.3923/itj.2008.1030.1036.
6
Akgun, T., Y. Altunbasak, and R. M. Mersereau. "Super-resolution reconstruction of hyperspectral images." IEEE Transactions on Image Processing 14, no. 11 (November 2005): 1860–75. http://dx.doi.org/10.1109/tip.2005.854479.
7
Bu, Lijing, Dong Dai, Zhengpeng Zhang, Yin Yang, and Mingjun Deng. "Hyperspectral Super-Resolution Reconstruction Network Based on Hybrid Convolution and Spectral Symmetry Preservation." Remote Sensing 15, no. 13 (June 21, 2023): 3225. http://dx.doi.org/10.3390/rs15133225.
Abstract:
Hyperspectral images (HSI) have high-dimensional and complex spectral characteristics, with dozens or even hundreds of bands covering the same area of pixels. The rich information of the ground objects makes hyperspectral images widely used in satellite remote sensing. Due to the limitations of remote sensing satellite sensors, hyperspectral images suffer from insufficient spatial resolution. Therefore, utilizing software algorithms to improve the spatial resolution of hyperspectral images has become an urgent problem that needs to be solved. The spatial information and spectral information of hyperspectral images are strongly correlated. If only the spatial resolution is improved, it often damages the spectral information. Inspired by the high correlation between spectral information in adjacent spectral bands of hyperspectral images, a hybrid convolution and spectral symmetry preservation network has been proposed for hyperspectral super-resolution reconstruction. This includes a model to integrate information from neighboring spectral bands to supplement target band feature information. The proposed model introduces flexible spatial-spectral symmetric 3D convolution in the network structure to extract low-resolution and neighboring band features. At the same time, a combination of deformable convolution and attention mechanisms is used to extract information from low-resolution bands. Finally, multiple bands are fused in the reconstruction module, and the high-resolution hyperspectral image containing global information is obtained by Fourier transform upsampling. Experiments were conducted on the indoor hyperspectral image dataset CAVE, the airborne hyperspectral dataset Pavia Center, and Chikusei. In the X2 super-resolution task, the PSNR values achieved on the CAVE, Pavia Center, and Chikusei datasets were 46.335, 36.321, and 46.310, respectively. In the X4 super-resolution task, the PSNR values achieved on the CAVE, Pavia Center, and Chikusei datasets were 41.218, 30.377, and 38.365, respectively. The results show that our method outperforms many advanced algorithms in objective indicators such as PSNR and SSIM while maintaining the spectral characteristics of hyperspectral images.
8
He, Zhi, and Lin Liu. "Hyperspectral Image Super-Resolution Inspired by Deep Laplacian Pyramid Network." Remote Sensing 10, no. 12 (December 2, 2018): 1939. http://dx.doi.org/10.3390/rs10121939.
Abstract:
Existing hyperspectral sensors usually produce high-spectral-resolution but low-spatial-resolution images, and super-resolution has yielded impressive results in improving the resolution of the hyperspectral images (HSIs). However, most of the super-resolution methods require multiple observations of the same scene and improve the spatial resolution without fully considering the spectral information. In this paper, we propose an HSI super-resolution method inspired by the deep Laplacian pyramid network (LPN). First, the spatial resolution is enhanced by an LPN, which can exploit the knowledge from natural images without using any auxiliary observations. The LPN progressively reconstructs the high-spatial-resolution images in a coarse-to-fine fashion by using multiple pyramid levels. Second, spectral characteristics between the low- and high-resolution HSIs are studied by the non-negative dictionary learning (NDL), which is proposed to learn the common dictionary with non-negative constraints. The super-resolution results can finally be obtained by multiplying the learned dictionary and its corresponding sparse codes. Experimental results on three hyperspectral datasets demonstrate the feasibility of the proposed method in enhancing the spatial resolution of the HSI with preserving the spectral information simultaneously.
9
Cao, Meng, Wenxing Bao, and Kewen Qu. "Hyperspectral Super-Resolution Via Joint Regularization of Low-Rank Tensor Decomposition." Remote Sensing 13, no. 20 (October 14, 2021): 4116. http://dx.doi.org/10.3390/rs13204116.
Abstract:
The hyperspectral image super-resolution (HSI-SR) problem aims at reconstructing the high resolution spatial–spectral information of the scene by fusing low-resolution hyperspectral images (LR-HSI) and the corresponding high-resolution multispectral image (HR-MSI). In order to effectively preserve the spatial and spectral structure of hyperspectral images, a new joint regularized low-rank tensor decomposition method (JRLTD) is proposed for HSI-SR. This model alleviates the problem that the traditional HSI-SR method, based on tensor decomposition, fails to adequately take into account the manifold structure of high-dimensional HR-HSI and is sensitive to outliers and noise. The model first operates on the hyperspectral data using the classical Tucker decomposition to transform the hyperspectral data into the form of a three-mode dictionary multiplied by the core tensor, after which the graph regularization and unidirectional total variational (TV) regularization are introduced to constrain the three-mode dictionary. In addition, we impose the l1-norm on core tensor to characterize the sparsity. While effectively preserving the spatial and spectral structures in the fused hyperspectral images, the presence of anomalous noise values in the images is reduced. In this paper, the hyperspectral image super-resolution problem is transformed into a joint regularization optimization problem based on tensor decomposition and solved by a hybrid framework between the alternating direction multiplier method (ADMM) and the proximal alternate optimization (PAO) algorithm. Experimental results conducted on two benchmark datasets and one real dataset show that JRLTD shows superior performance over state-of-the-art hyperspectral super-resolution algorithms.
10
Li, Xiaoyan, Lefei Zhang, and Jane You. "Domain Transfer Learning for Hyperspectral Image Super-Resolution." Remote Sensing 11, no. 6 (March 22, 2019): 694. http://dx.doi.org/10.3390/rs11060694.
Abstract:
A Hyperspectral Image (HSI) contains a great number of spectral bands for each pixel; however, the spatial resolution of HSI is low. Hyperspectral image super-resolution is effective to enhance the spatial resolution while preserving the high-spectral-resolution by software techniques. Recently, the existing methods have been presented to fuse HSI and Multispectral Images (MSI) by assuming that the MSI of the same scene is required with the observed HSI, which limits the super-resolution reconstruction quality. In this paper, a new framework based on domain transfer learning for HSI super-resolution is proposed to enhance the spatial resolution of HSI by learning the knowledge from the general purpose optical images (natural scene images) and exploiting the cross-correlation between the observed low-resolution HSI and high-resolution MSI. First, the relationship between low- and high-resolution images is learned by a single convolutional super-resolution network and then is transferred to HSI by the idea of transfer learning. Second, the obtained Pre-high-resolution HSI (pre-HSI), the observed low-resolution HSI, and high-resolution MSI are simultaneously considered to estimate the endmember matrix and the abundance code for learning the spectral characteristic. Experimental results on ground-based and remote sensing datasets demonstrate that the proposed method achieves comparable performance and outperforms the existing HSI super-resolution methods.
Dissertations / Theses on the topic "Hyperspectral super-Resolution":
1
Abi, rizk Ralph. "High-resolution hyperspectral reconstruction by inversion of integral field spectroscopy measurements. Application to the MIRI-MRS infrared spectrometer of the James Webb Space Telescope." Electronic Thesis or Diss., université Paris-Saclay, 2021. http://www.theses.fr/2021UPASG087.
Abstract:
Cette thèse traite des approches de type problème inverse pour reconstruire une image 3D spatio-spectrale (2D+λ) à partir d’un ensemble de mesures infrarouges 2D fournies par l’instrument “Integral Field Spectrometer” (IFS) (Mid-ResolutionSpectrometer: MRS) de l’instrument “Mid-Infrared” à bord du “James Webb Space Telescope” (JWST). Plusieurs difficultées se posent lors de la reconstruction car l’instrument IFS contient des composantes complexes qui dégradent et modifient les mesures: (1) les réponses des composantes ne sont pas parfaites et introduisent un flou spatial et spectral aux mesures qui dépendent de la longueur d’onde, (2) l’instrument considère plusieurs observations avec plusieurs champs de vue (comme les canaux spectraux et les fentes parallèles), (3) les sorties de l’instrument sont projetées sur plusieurs détecteurs 2D et échantillonnées avec des pas d’échantillonnage hétérogènes.La reconstruction d’image 2D+λ est un problème mal posé principalement en raison du flou spatio-spectral et de l’échantillonnage spatial insuffisant. Pour compenser la perte d’informations spatiales, le MRS permet des observations multiples de la même scène d’entrée en décalant le pointage du télescope, conduisant à un problème de Super-Résolution (SR). Nous proposons un algorithme de reconstruction qui traite conjointement les informations spatiales et spectrales des mesures 2D suivant deux étapes. Tout d’abord, nous concevons un modèle direct qui décrit la réponse des composantes de l’instrument IFS comme une série d’opérateurs mathématiques et qui établit une relation entre les mesures et l’image 2D+λ d’entrée qu’on cherche à reconstruire. Ensuite, le modèle direct est utilisé pour reconstruire l’image 2D+λ en s’appuyant sur l’approche des moindres carrés régularisée avec une régularisation convexe pour la préservation des contours. Nous nous appuyons sur les approches semi quadratiques rapides basées sur la formulation de Geman et Reynolds pour résoudre le problème. L’algorithme de reconstruction proposé comprend principalement une étape de fusion des mesures issues de différentes observations spatio-spectrales avec différents flous et différents échantillonnages, une étape de SR à partir des différents pointages de l’instrument, et une étape de déconvolution pour minimiser le flou. Un autre modèle direct pour le même instrument est également développé dans notre travail, en supposant que l’image 2D+λ d’entrée vit dans un sous-espace de faible dimension et peut être modélisée comme une combinaison linéaire de composantes spectrales, supposées connues, pondérées par des coefficients de mélange inconnus. Nous nous appuyons ensuite sur l’algorithme d’optimisation Majorize-Minimize Memory Gradient (3MG) pour estimer les coefficients de mélange inconnus. L’approximation par sous-espace réduit le nombre d’inconnues. Par conséquent, le rapport signal sur bruit augmente. De plus, le modèle de mélange de source avec des composantes spectrales connues permet de conserver l’information spectrale complexe de l’image 2D+λ reconstruite. La reconstruction proposée est testée sur plusieurs images 2D+λ synthétiques ayant des différentes distributions spatiales et spectrales. Notre reconstruction montre une déconvolution nette et une amélioration significative des résolutions spatiales et spectrales des images 2D+λ reconstruites par rapport aux algorithmes de l’état de l’art, notamment autour des bordsThis thesis deals with inverse problem approaches to reconstruct a 3D spatio-spectral image from a set of 2D infrared measurements provided by the Integral Field Spectrometer (IFS) instrument (Mid-Resolution Spectrometer: MRS) of the Mid-Infrared Instrument onboard the James Webb Space Telescope. The reconstruction is challenging because the IFS involves complex components that degrade the measurements: (1) the responses of the components are not perfect and introduce a wavelength-dependent spatial and spectral blurring, (2) the instrument considers several observations of the input with several spatial and spectral fields of views, (3) the output measurements are projected onto multiple 2D detectors and sampled with heterogeneous step sizes. The 3D image reconstruction is an ill-posed problem mainly due to spatio-spectral blurring and insufficient spatial sampling. To compensate for the loss of spatial information, the MRS allows multiple observations of the same scene by shifting the telescope pointing, leading to a multi-frame Super-Resolution (SR) problem. We propose an SR reconstruction algorithm that jointly processes the spatial and spectral information of the degraded 2D measurements following two main steps. First, we design a forward model that describes the response of the IFS instrument as a series of mathematical operators and establishes a relationship between the measurements and the unknown 3D input image. Next, the forward model is used to reconstruct the unknown input.The reconstruction is based on the regularized least square approach with a convex regularization for edge-preserving. We rely on the fast half-quadratic approaches based on Geman and Reynolds formulation to solve the problem. The proposed algorithm mainly includes a fusion step of measurements from different spatio-spectral observations with different blur and different sampling, a multi-frame Super-Resolution step from the different pointing of the instrument, and a deconvolution step to minimize the blurring. Another forward model for the same instrument is also developed in our work, by assuming that the 3D input image lives in a low dimensional subspace and can be modeled as a linear combination of spectral components, assumed known, weighted by unknown mixing coefficients, known as the Linear Mixing Model (LMM). We then rely on the Majorize-Minimize Memory Gradient (3MG) optimization algorithm to estimate the unknown mixing coefficients. The subspace approximation reduces the number of the unknowns. Consequently, the signal-to-noise ratio is increased. In addition, the LMM formulation with known spectral components allows preserving the complex spectral information of the reconstructed 3D image. The proposed reconstruction is tested on several synthetic HS images with different spatial and spectral distributions. Our algorithm shows a clear deconvolution and a significant improvement of the spatial and spectral resolutions of the reconstructed images compared to the state-of-art algorithms, particularly around the edges
2
Wei, Qi. "Bayesian fusion of multi-band images : A powerful tool for super-resolution." Phd thesis, Toulouse, INPT, 2015. http://oatao.univ-toulouse.fr/14398/1/wei.pdf.
Abstract:
Hyperspectral (HS) imaging, which consists of acquiring a same scene in several hundreds of contiguous spectral bands (a three dimensional data cube), has opened a new range of relevant applications, such as target detection [MS02], classification [C.-03] and spectral unmixing [BDPD+12]. However, while HS sensors provide abundant spectral information, their spatial resolution is generally more limited. Thus, fusing the HS image with other highly resolved images of the same scene, such as multispectral (MS) or panchromatic (PAN) images is an interesting problem. The problem of fusing a high spectral and low spatial resolution image with an auxiliary image of higher spatial but lower spectral resolution, also known as multi-resolution image fusion, has been explored for many years [AMV+11]. From an application point of view, this problem is also important as motivated by recent national programs, e.g., the Japanese next-generation space-borne hyperspectral image suite (HISUI), which fuses co-registered MS and HS images acquired over the same scene under the same conditions [YI13]. Bayesian fusion allows for an intuitive interpretation of the fusion process via the posterior distribution. Since the fusion problem is usually ill-posed, the Bayesian methodology offers a convenient way to regularize the problem by defining appropriate prior distribution for the scene of interest. The aim of this thesis is to study new multi-band image fusion algorithms to enhance the resolution of hyperspectral image. In the first chapter, a hierarchical Bayesian framework is proposed for multi-band image fusion by incorporating forward model, statistical assumptions and Gaussian prior for the target image to be restored. To derive Bayesian estimators associated with the resulting posterior distribution, two algorithms based on Monte Carlo sampling and optimization strategy have been developed. In the second chapter, a sparse regularization using dictionaries learned from the observed images is introduced as an alternative of the naive Gaussian prior proposed in Chapter 1. instead of Gaussian prior is introduced to regularize the ill-posed problem. Identifying the supports jointly with the dictionaries circumvented the difficulty inherent to sparse coding. To minimize the target function, an alternate optimization algorithm has been designed, which accelerates the fusion process magnificently comparing with the simulation-based method. In the third chapter, by exploiting intrinsic properties of the blurring and downsampling matrices, a much more efficient fusion method is proposed thanks to a closed-form solution for the Sylvester matrix equation associated with maximizing the likelihood. The proposed solution can be embedded into an alternating direction method of multipliers or a block coordinate descent method to incorporate different priors or hyper-priors for the fusion problem, allowing for Bayesian estimators. In the last chapter, a joint multi-band image fusion and unmixing scheme is proposed by combining the well admitted linear spectral mixture model and the forward model. The joint fusion and unmixing problem is solved in an alternating optimization framework, mainly consisting of solving a Sylvester equation and projecting onto a simplex resulting from the non-negativity and sum-to-one constraints. The simulation results conducted on synthetic and semi-synthetic images illustrate the advantages of the developed Bayesian estimators, both qualitatively and quantitatively.
3
Prévost, Clémence. "Multimodal data fusion by coupled low-rank tensor approximations." Electronic Thesis or Diss., Université de Lorraine, 2021. http://www.theses.fr/2021LORR0180.
Abstract:
Grâce au développement de nouvelles modalités, de plus en plus de signaux sont collectés chaque jour. Ainsi, il est fréquent que différents signaux renferment des informations sur un même phénomène physique. Cependant, un seul signal peut ne contenir que des informations partielles, d’où l’émergence de la fusion de données multimodales pour dépasser cette limitation. La fusion de données est définie comme le traitement conjoint de signaux issus de différentes modalités. Son but est d’exploiter à plein les capacités de chaque modalité à extraire du savoir sur le phénomène d’intérêt, tout en mettant en lumière des informations supplémentaires issues de la fusion. Cependant, dû aux interactions complexes entre les différentes modalités, dresser un tableau des avantages des modèles de fusion et de leurs limites par rapport au traitement séparé est une tâche complexe. Dans certains domaines tels que l’imagerie biomédicale ou la télédétection, les signaux observés sont des cubes de données appelés tenseurs ; ainsi, il est possible d’envisager des modèles de fusion tensorielle. En effet, la modélisation tensorielle de rang faible permet de préserver la structure des observations tout en jouissant des propriétés d’unicité des décompositions de tenseurs. Dans cette thèse, on s’intéresse à un problème de reconstruction d'un tenseur à haute résolution à partir d’observations tensorielles faiblement résolues. En particulier, le problème de super-résolution hyperspectrale (HSR) vise à reconstruire un tenseur à partir de deux versions dégradées : tandis que l’une est faiblement résolue dans deux modes spatiaux, la seconde est faiblement résolue dans le troisième mode spectral. Des approches tensorielles ont été récemment proposées, sous l’hypothèse d’une décomposition tensorielle de rang faible du tenseur à haute résolution. Les premiers travaux à exploiter cette hypothèse se basent sur la décomposition canonique polyadique (CP) et ont donné lieu à de nombreuses méthodes tensorielles de reconstruction, incluant ce travail. La première partie de cette thèse est dédiée au développement d’algorithmes tensoriels pour le problème HSR. Dans le Chapitre 2, nous proposons une reformulation sous forme d’une décomposition de Tucker couplée, ainsi que deux algorithmes analytiques basés sur la décomposition en valeurs singulières d’ordre supérieur. Les simulations illustrent des performances compétitives au regard des méthodes de l'état de l'art, avec un temps de calcul réduit. Le Chapitre 3 utilise un modèle de variabilité spectrale. Le problème de reconstruction est reformulé grâce à une décomposition bloc-termes. Les facteurs de la décomposition sont contraints à être positifs afin de garantir leur interprétabilité physique dans un modèle de mélange. Ainsi, cette approche propose une solution conjointe au problème HSR et au problème de démélange spectral. La seconde partie de cette thèse consiste en l’étude des performances statistiques des modèles tensoriels couplés. Cette partie vise à évaluer l’efficacité de certains algorithmes présentés à la première partie. Dans le Chapitre 4, on considère les bornes de Cramér-Rao sous contraintes (CCRB) pour des modèlesCP couplés. L’expression de la matrice d’information de Fisher est fournie dans deux scénarios, selon que i) l’on considère le problème de reconstruction totalement couplé seulement, ou ii) l’on cherche à comparer les performances des modèles totalement couplé, partiellement couplé et découplé. L’efficacité asymptotique des algorithmes CP existants est également illustrée.Le Chapitre 5 considère un problème d’estimation non-standard dans lequel les contraintes sur les paramètres déterministes du modèle impliquent un paramètre aléatoire. Dans ce contexte, la CCRB standard est non-informative. De fait, on introduit une nouvelle borne de Cramér-Rao sous contraintes aléatoires (RCCRB). Son intérêt est illustré au moyen d’un modèle bloc-termes couplé avec incertitudesDue to the recent emergence of new modalities, the amount of signals collected daily has been increasing. As a result, it frequently occurs that various signals provide information about the same phenomenon. However, a single signal may only contain partial information about this phenomenon. Multimodal data fusion was proposed to overcome this issue. It is defined as joint processing of datasets acquired from different modalities. The aim of data fusion is to enhance the capabilities of each modality to express their specific information about the phenomenon of interest; it is also expected from data fusion that it brings out additional information that would be ignored by separate processing. However, due to the complex interactions between the modalities, understanding the advantages and limits of data fusion may not be straightforward.In a lot of applications such as biomedical imaging or remote sensing, the observed signals are three-dimensional arrays called tensors, thus tensor-based data fusion can be envisioned. Tensor low-rank modeling preserves the multidimensional structure of the observations and enjoys interesting uniqueness properties arising from tensor decompositions. In this work, we address the problem of recovering a high-resolution tensor from tensor observations with some lower resolutions.In particular, hyperspectral super-resolution (HSR) aims at reconstructing a tensor from two degraded versions. While one is degraded in two (spatial) modes, the second is degraded in the third (spectral) mode. Recently, tensor-based approaches were proposed for solving the problem at hand. These works are based on the assumption that the target tensor admits a given low-rank tensor decomposition. The first work addressing the problem of tensor-based HSR was based on a coupled canonical polyadic (CP) decomposition of the observations. This approach gave rise to numerous following reconstruction methods based on coupled tensor models, including our work.The first part of this thesis is devoted to the design of tensor-based algorithms for solving the HSR problem. In Chapter 2, we propose to formulate the problem as a coupled Tucker decomposition. We introduce two simple but fast algorithms based on the higher-order singular value decomposition of the observations. Our experiments show that our algorithms have a competitive performance with state-of-the-art tensor and matrix methods, with a lower computational time. In Chapter 3, we consider spectral variability between the observations. We formulate the reconstruction problem as a coupled block-term decomposition. We impose non-negativity of the low-rank factors, so that they can be incorporated into a physically-informed mixing model. Thus the proposed approach provides a solution to the joint HSR and unmixing problems.The second part of this thesis adresses the performance analysis of the coupled tensor models. The aim of this part is to assess the efficiency of some algorithms introduced in the first part. In Chapter 4, we consider constrained Cramér-Rao lower bounds (CCRB) for coupled tensor CP models. We provide a closed-form expression for the constrained Fisher information matrix in two scenarios, whether i) we only consider the fully-coupled reconstruction problem or ii) if we are interested in comparing the performance of fully-coupled, partially-coupled and uncoupled approaches. We prove that the existing CP-based algorithms are asymptotically efficient. Chapter 5 addresses a non-standard estimation problem in which the constraints on the deterministic model parameters involve a random parameter. We show that in this case, the standard CCRB is a non-informative bound. As a result, we introduce a new randomly constrained Cramér-Rao bound (RCCRB). The relevance of the RCCRB is illustrated using a coupled block-term decomposition model accounting for random uncertainties
4
Hugelier, Siewert. "Approaches to inverse problems in chemical imaging : applications in super-resolution and spectral unmixing." Thesis, Lille 1, 2017. http://www.theses.fr/2017LIL10144/document.
Abstract:
L’imagerie chimique permet d’accéder à la distribution spatiale des espèces chimiques. Nous distinguerons dans cette thèse deux types d’images différents: les images spatiales-temporelles et les images spatiales-spectrales.La microscopie de fluorescence super-résolue a commencé avec un faible nombre de fluorophores actifs par image. Actuellement, ça a évolué vers l’imagerie en haute densité qui requiert de nouvelles façons d’analyse. Nous proposons SPIDER, une approche de déconvolution par moindres carrés pénalisés. La considération de plusieurs pénalités permet de traduire les propriétés des émetteurs utilisés dans l'imagerie de fluorescence super-résolue. L'utilisation de cette méthode permet d'étudier des changements structuraux et morphologiques dans les échantillons biologiques. La méthode a été appliquée à l’imagerie sur cellules vivantes d’une cellule HEK-293T encodée par la protéine fluorescente DAKAP-Dronpa. On a pu obtenir une résolution spatiale de 55nm pour un temps d’acquisition de 0.5s.La résolution d'images hyperspectrales avec MCR-ALS fournit des informations spatiales et spectrales des contributions individuelles dans le mélange. Néanmoins, le voisinage des pixels est perdu du fait du dépliement du cube de données hyperspectrales sous forme d’une matrice bidirectionnelle. L’implémentation de contraintes spatiales n’est donc pas possible en MCR-ALS. Nous proposons une approche alternative dans laquelle une étape de repliement/dépliement est effectuée à chaque itération qui permet d’ajouter des fonctionnalités spatiales globales à la palette des contraintes. Nous avons développé plusieurs contraintes et on montre leur application aux données expérimentalesBesides the chemical information, chemical imaging also offers insights in the spatial distribution of the samples. Within this thesis, we distinguish between two different types of images: spatial-temporal images (super-resolution fluorescence microscopy) and spatial-spectral images (unmixing). In early super-resolution fluorescence microscopy, a low number of fluorophores were active per image. Currently, the field evolves towards high-density imaging that requires new ways of analysis. We propose SPIDER, an image deconvolution approach with multiple penalties. These penalties directly translate the properties of the blinking emitters used in super-resolution fluorescence microscopy imaging. SPIDER allows investigating highly dynamic structural and morphological changes in biological samples with a high fluorophore density. We applied the method on live-cell imaging of a HEK-293T cell labeled with DAKAP-Dronpa and demonstrated a spatial resolution down to 55 nm and a time sampling of 0.5 s. Unmixing hyperspectral images with MCR-ALS provides spatial and spectral information of the individual contributions in the mixture. Due to loss of the pixel neighborhood during the unfolding of the hyperspectral data cube to a two-way matrix, spatial information cannot be added as a constraint during the analysis We therefore propose an alternative approach in which an additional refolding/unfolding step is performed in each iteration. This data manipulation allows global spatial features to be added to the palette of MCR-ALS constraints. From this idea, we also developed several constraints and show their application on experimental data
5
Raynaud, Christophe. "Spectroscopie d'absorption et d'émission des excitons dans les nanotubes de carbone." Thesis, Sorbonne Paris Cité, 2018. http://www.theses.fr/2018USPCC199/document.
Abstract:
Les propriétés optiques de nanotubes de carbone sont décrites idéalement parla physique d’un objet unidimensionnel, donnant lieu notamment à l’apparition des excitons pour décrire les transitions optiques de ces objets. Les expériences d’optique(émission, absorption) réalisées sur ces objets à température ambiante et sur des ensemble d’objets ont permis de confirmer les prédictions théoriques basées sur la physique des objets 1D. Mais à température cryogénique et à l’échelle de l’objet unique,les propriétés optiques observées expérimentalement sont systématiquement très éloignées de celles d’un objet 1D. On peut notamment citer l’apparition de propriétés comme l’émission de photons uniques, qui a largement contribué à l’intensification de la recherche sur ces objets pour des applications en photonique quantique. Ces propriétés sont attribuées à la localisation des excitons le long de l’axe des nanotubes dans des puits de potentiel créés aléatoirement par l’interaction des nanotubes avec leur environnement. Les propriétés optiques sont alors proches de celles des objets0D, et sont fortement modulées par l’environnement. Les mécanismes et l’origine de la localisation et la connaissance physique de ces puits sont encore très limités. Ce travail montre d’une part le développement d’une technique d’absorption sur objet individuel et la caractérisation de sa sensibilité, et d’autre part l’étude statistique de l’émission de nanotubes à température cryogénique. Les résultats obtenus par une technique de super-résolution couplée à une imagerie hyper-spectrale montrent les grandeurs caractéristiques des puits de potentiels au sein de nanotubes individuels.Un dispositif expérimental de photoluminescence résolue en excitation implémenté au cours de ce travail a également montré une modification de l’état excitonique fondamental par l’environnement, avec l’apparition d’une discrétisation spatiale et spectrale de l’état fondamental délocalisé en une multitude d’états localisésThe optical properties of carbon nanotubes are ideally described by the physicsof a one-dimensional object, giving rise in particular to the emergence of excitons todescribe the optical transitions of these objects. The optical experiments (emission,absorption) carried out on these objects at ambient temperature and on ensemblesconfirm the theoretical predictions based on the physics of 1D objects. But atcryogenic temperature and at the single emitter scale, the optical properties observedexperimentally are systematically different from those of a 1D object. One can citethe emergence of properties such as photon antibunching, which largely contributed tothe intensification of research on these objects for applications in quantum photonics.These properties are attributed to the localization of excitons along the nanotube axisin local potential wells (traps) created randomly by the interaction of nanotubes withtheir environment. The optical properties are then close to those of 0D objects, andare strongly modulated by the environment. The mechanisms and the origin of thelocalization and the physical knowledge of these traps are still very limited. This workshows on the one hand the development of an absorption setup on individual objectand the characterization of its sensitivity, and on the other hand the statistical studyof the emission of nanotubes at cryogenic temperature in a micro-photoluminescencesetup. The results obtained in the later setup by a super-resolution technique coupledwith hyper-spectral imaging show the characteristic quantities of potential wellswithin individual nanotubes. An experimental excitation-resolved photoluminescencesetup implemented during this work also showed a modification of the fundamentalexcitonic state by the environment, with the emergence of a spatial and spectraldiscretization of the delocalized ground state in a multitude of localized states
6
Hsieh, Chih-Hsiang, and 謝智翔. "A Convex Optimization Based Coupled Non-negative Matrix Factorization Algorithm for Hyperspectral Image Super-resolution." Thesis, 2017. http://ndltd.ncl.edu.tw/handle/megg23.
Abstract:
碩士國立清華大學
通訊工程研究所
106
In recent years, fusing a low-spatial-resolution hyperspectral image with a highspatial-resolution multispectral image has been thought of as an economical approach for obtaining high-spatial-resolution hyperspectral image. A fusion criterion, termed coupled nonnegative matrix factorization (CNMF) has been reported to be effective in yielding promising fusion performance. However, the CNMF criterion amounts to an ill-posed inverse problem. In this thesis, we propose a new data fusion algorithm by suitable regularization that significantly outperforms the unregularized CNMF algorithm. Besides utilizing the sparsity-promoting regularizer, which promotes the sparsity of the abundance map, we also incorporate the sum-of-squared endmember distances demoting regularizer. Owing to the bi-convexity of the formulated optimization problem, we can decouple it into two convex subproblems. Each subproblem is then solved by a carefully designed alternating direction method of multiplers (ADMM), leading to a convex-optimization based CNMF (CO-CNMF) fusion algorithm, where each ADMM iterate is equipped with a closed-form solution. Since the problem size is very large, leading to high computational complexity of the proposed CO-CNMF algorithm, we futher obtain alternative expressions by exploiting some inherent matrix structure in those closed-form solutions, which greatly reduce the computational complexity. Finally, we present some experiments using real hyperspectral data, which can be divided into three parts. The first part is to analyze how we choose parameters in our proposed algorithm. Second, we demonstrate its superior performance to some state-of-the-art fusion algorithms. Third, by some experimental results, we discuss its performance loss due to imperfect co-registration.
Book chapters on the topic "Hyperspectral super-Resolution":
1
Wang, Liguo, and Chunhui Zhao. "Super-Resolution Technique of HSI." In Hyperspectral Image Processing, 187–216. Berlin, Heidelberg: Springer Berlin Heidelberg, 2015. http://dx.doi.org/10.1007/978-3-662-47456-3_6.
2
Sudheer Babu, R., and K. E. Sreenivasa Murthy. "Enhanced Joint Estimation-Based Hyperspectral Image Super Resolution." In Advances in Intelligent Systems and Computing, 503–16. Singapore: Springer Singapore, 2018. http://dx.doi.org/10.1007/978-981-10-7868-2_49.
3
Bu, Yuanyang, Yongqiang Zhao, and Jonathan Cheung-Wai Chan. "Hyperspectral Image Super-Resolution via Self-projected Smooth Prior." In Pattern Recognition and Computer Vision, 648–59. Cham: Springer International Publishing, 2020. http://dx.doi.org/10.1007/978-3-030-60633-6_54.
4
Akhtar, Naveed, Faisal Shafait, and Ajmal Mian. "Sparse Spatio-spectral Representation for Hyperspectral Image Super-resolution." In Computer Vision – ECCV 2014, 63–78. Cham: Springer International Publishing, 2014. http://dx.doi.org/10.1007/978-3-319-10584-0_5.
5
Bu, Yuanyang, Yongqiang Zhao, Jize Xue, and Jonathan Cheung-Wai Chan. "Smooth Coupled Tucker Decomposition for Hyperspectral Image Super-Resolution." In Pattern Recognition and Computer Vision, 238–48. Cham: Springer International Publishing, 2021. http://dx.doi.org/10.1007/978-3-030-88010-1_20.
6
Sun, He, Zhiwei Zhong, Deming Zhai, Xianming Liu, and Junjun Jiang. "Hyperspectral Image Super-Resolution Using Multi-scale Feature Pyramid Network." In Communications in Computer and Information Science, 49–61. Singapore: Springer Singapore, 2020. http://dx.doi.org/10.1007/978-981-15-3341-9_5.
7
Wang, Chen, Yun Liu, Xiao Bai, Wenzhong Tang, Peng Lei, and Jun Zhou. "Deep Residual Convolutional Neural Network for Hyperspectral Image Super-Resolution." In Lecture Notes in Computer Science, 370–80. Cham: Springer International Publishing, 2017. http://dx.doi.org/10.1007/978-3-319-71598-8_33.
8
Jia, Huidi, Siyu Guo, Zhenyu Li, Xi’ai Chen, Zhi Han, and Yandong Tang. "Low-Rank Tensor Tucker Decomposition for Hyperspectral Images Super-Resolution." In Intelligent Robotics and Applications, 502–12. Cham: Springer International Publishing, 2022. http://dx.doi.org/10.1007/978-3-031-13822-5_45.
9
Zhang, Jing, Zekang Wan, Minhao Shao, and Yunsong Li. "A Multi-path Neural Network for Hyperspectral Image Super-Resolution." In Lecture Notes in Computer Science, 377–87. Cham: Springer International Publishing, 2021. http://dx.doi.org/10.1007/978-3-030-87361-5_31.
10
Yao, Jing, Danfeng Hong, Jocelyn Chanussot, Deyu Meng, Xiaoxiang Zhu, and Zongben Xu. "Cross-Attention in Coupled Unmixing Nets for Unsupervised Hyperspectral Super-Resolution." In Computer Vision – ECCV 2020, 208–24. Cham: Springer International Publishing, 2020. http://dx.doi.org/10.1007/978-3-030-58526-6_13.
Conference papers on the topic "Hyperspectral super-Resolution":
1
Chanussot, Jocelyn. "On Hyperspectral Super-Resolution." In IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2021. http://dx.doi.org/10.1109/igarss47720.2021.9553903.
2
Yongqiang Zhao, Chen Yi, Jingxiang Yang, and Jonathan Cheung-Wai Chan. "Coupled hyperspectral super-resolution and unmixing." In IGARSS 2014 - 2014 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2014. http://dx.doi.org/10.1109/igarss.2014.6947016.
3
Elbakary, Mohamed, and Mohammad S. Alam. "Super-resolution reconstruction of hyperspectral images." In Defense and Security Symposium, edited by David P. Casasent and Tien-Hsin Chao. SPIE, 2007. http://dx.doi.org/10.1117/12.718383.
4
Nie, Jiangtao, Lei Zhang, Cong Wang, Wei Wei, and Yanning Zhang. "Robust Deep Hyperspectral Imagery Super-Resolution." In IGARSS 2019 - 2019 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2019. http://dx.doi.org/10.1109/igarss.2019.8900117.
5
Hussain, Sadia, and Brejesh Lall. "Spectral Grouping Driven Hyperspectral Super-Resolution." In 2023 IEEE International Conference on Image Processing (ICIP). IEEE, 2023. http://dx.doi.org/10.1109/icip49359.2023.10222288.
6
Yuan, Han, Fengxia Yan, Xinmeng Chen, and Jubo Zhu. "Compressive Hyperspectral Imaging and Super-resolution." In 2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC). IEEE, 2018. http://dx.doi.org/10.1109/icivc.2018.8492822.
7
Irmak, Hasan, Gozde Bozdagi Akar, and Seniha Esen Y. uksel. "Image Fusion for Hyperspectral Image Super-Resolution." In 2018 9th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS). IEEE, 2018. http://dx.doi.org/10.1109/whispers.2018.8747231.
8
Lanaras, Charis, Emmanuel Baltsavias, and Konrad Schindler. "Hyperspectral Super-Resolution by Coupled Spectral Unmixing." In 2015 IEEE International Conference on Computer Vision (ICCV). IEEE, 2015. http://dx.doi.org/10.1109/iccv.2015.409.
9
Li, Ke, Dengxin Dai, and Luc Van Gool. "Hyperspectral Image Super-Resolution with RGB Image Super-Resolution as an Auxiliary Task." In 2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). IEEE, 2022. http://dx.doi.org/10.1109/wacv51458.2022.00409.
10
Aburaed, Nour, Mohammed Alkhatib, Stephen Marshall, Jaime Zabalza, and Hussain Al-Ahmad. "Complex-valued neural network for hyperspectral single image super resolution." In Hyperspectral Imaging and Applications II, edited by Nick J. Barnett, Aoife A. Gowen, and Haida Liang. SPIE, 2023. http://dx.doi.org/10.1117/12.2645086.