Centre dedocumentation
scientifique

Unsupervised multi-level feature extraction for improvement of hyperspectral classification / Qiaoqiao Sun in Remote sensing, vol 13 n° 8 (April-2 2021)  

Public [article]  inRemote sensing > vol 13 n° 8 (April-2 2021) . - n° 1602 Titre : Unsupervised multi-level feature extraction for improvement of hyperspectral classification Type de document : Article/Communication Auteurs : Qiaoqiao Sun, Auteur ; Xuefeng Liu, Auteur ; Salah Bourennane, Auteur Année de publication : 2021 Article en page(s) : n° 1602 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage profond [Termes IGN] classification non dirigée [Termes IGN] codage [Termes IGN] convolution (signal) [Termes IGN] déconvolution [Termes IGN] échantillonnage d'image [Termes IGN] extraction de traits caractéristiques [Termes IGN] image hyperspectrale [Termes IGN] observation multiniveaux Résumé : (auteur) Deep learning models have strong abilities in learning features and they have been successfully applied in hyperspectral images (HSIs). However, the training of most deep learning models requires labeled samples and the collection of labeled samples are labor-consuming in HSI. In addition, single-level features from a single layer are usually considered, which may result in the loss of some important information. Using multiple networks to obtain multi-level features is a solution, but at the cost of longer training time and computational complexity. To solve these problems, a novel unsupervised multi-level feature extraction framework that is based on a three dimensional convolutional autoencoder (3D-CAE) is proposed in this paper. The designed 3D-CAE is stacked by fully 3D convolutional layers and 3D deconvolutional layers, which allows for the spectral-spatial information of targets to be mined simultaneously. Besides, the 3D-CAE can be trained in an unsupervised way without involving labeled samples. Moreover, the multi-level features are directly obtained from the encoded layers with different scales and resolutions, which is more efficient than using multiple networks to get them. The effectiveness of the proposed multi-level features is verified on two hyperspectral data sets. The results demonstrate that the proposed method has great promise in unsupervised feature learning and can help us to further improve the hyperspectral classification when compared with single-level features. Numéro de notice : A2021-380 Affiliation des auteurs : non IGN Thématique : IMAGERIE/INFORMATIQUE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.3390/rs13081602 Date de publication en ligne : 20/04/2021 En ligne : https://doi.org/10.3390/rs13081602 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=97628 [article]

Applied signal processing / Sadasivan Puthusserypady (2021)  

Public Titre : Applied signal processing Type de document : Guide/Manuel Auteurs : Sadasivan Puthusserypady, Auteur Editeur : Boston, Delft : Now publishers Année de publication : 2021 Collection : *NowOpen* Importance : 550 p. ISBN/ISSN/EAN : 978-1-68083-979-1 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement du signal [Termes IGN] convolution (signal) [Termes IGN] filtrage du signal [Termes IGN] modulation d'amplitude [Termes IGN] modulation de fréquence [Termes IGN] série de Fourier [Termes IGN] signal aléatoire [Termes IGN] transformation de Fourier [Termes IGN] transformation de Hilbert Résumé : (éditeur) Being an inter-disciplinary subject, Signal Processing has application in almost all scientific fields. Applied Signal Processing tries to link between the analog and digital signal processing domains. Since the digital signal processing techniques have evolved from its analog counterpart, this book begins by explaining the fundamental concepts in analog signal processing and then progresses towards the digital signal processing. This will help the reader to gain a general overview of the whole subject and establish links between the various fundamental concepts. While the focus of this book is on the fundamentals of signal processing, the understanding of these topics greatly enhances the confident use as well as further development of the design and analysis of digital systems for various engineering and medical applications. Applied Signal Processing also prepares readers to further their knowledge in advanced topics within the field of signal processing. Note de contenu : 1- Introduction 2- Power and Energy 3- Fourier series 4- Fourier transform 5- Complex signals 6- Analog systems 7- Sampling and digital signals 8- Transform of discrete time signals 9- Fourier spectra of discrete-time signals 10- Digital systems 11- Implementation of digital systems 12- Discrete Fourier transform 13- Fast Fourier transform 14- Design of digital filters 15- Random signals 16- Modulation 17- Power Spectrum Estimation Numéro de notice : 28562 Affiliation des auteurs : non IGN Thématique : IMAGERIE/POSITIONNEMENT Nature : Manuel de cours DOI : 10.1561/9781680839791 En ligne : http://dx.doi.org/10.1561/9781680839791 Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=97593

Kalman filtering with state constraints applied to multi-sensor systems and georeferencing / Sören Vogel (2020)  

Public Titre : Kalman filtering with state constraints applied to multi-sensor systems and georeferencing Type de document : Thèse/HDR Auteurs : Sören Vogel, Auteur Editeur : Munich : Bayerische Akademie der Wissenschaften Année de publication : 2020 Collection : DGK - C, ISSN 0065-5325 num. 856 Importance : 144 p. ISBN/ISSN/EAN : 978-3-7696-5268-0 Note générale : bibliographie Diese Arbeit ist gleichzeitig veröffentlicht in: Wissenschaftliche Arbeiten der Fachrichtung Geodäsie und Geoinformatik der Universität Hannover ISSN 0174-1454, Nr. 364, 2020 Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Acquisition d'image(s) et de donnée(s) [Termes IGN] contrainte d'intégrité [Termes IGN] convolution (signal) [Termes IGN] étalonnage de capteur (imagerie) [Termes IGN] géoréférencement direct [Termes IGN] positionnement cinématique [Termes IGN] programmation par contraintes Résumé : (auteur) Active research on the development of autonomous vehicles has been carried out for several years now. However, some significant challenges still need to be solved in this context. Particularly relevant is the constant guarantee and assurance of the integrity of such autonomous systems. In order to ensure safe manoeuvring in the direct environment of humans, an accurate, precise, reliable and continuous determination of the vehicle’s position and orientation is mandatory. In geodesy, this process is also referred to as georeferencing with respect to a superordinate earth-fixed coordinate system. Especially for complex inner-city areas, there are no fully reliable methods available so far. The otherwise suitable and therefore common Global Navigation Satellite System (GNSS) observations can fail in urban canyons. However, this fact does not only apply exclusively to autonomous vehicles but can generally also be transferred to any kinematic Multi-Sensor System (MSS) operating within challenging environments. Especially in geodesy, there are many MSSs, which require accurate and reliable georeferencing regardless of the environment. This is indispensable for derived subsequent products, such as highly accurate three-dimensional point clouds for 3D city models or Building Information Modelling (BIM) applications. The demand for new georeferencing methods under aspects of integrity also involves the applicability of big data. Modern sensors for capturing the environment, e.g. laser scanners or cameras, are becoming increasingly cheaper and also offer higher information density and accuracy. For many kinematic MSSs, this change leads to a steady increase in the amount of acquired observation data. Many of the currently methods used are not suitable for processing such amounts of data, and instead, they only use a random subset. Besides, big data also influences potential requirements with regard to possible real-time applications. If there is no excessive computing power available to take into account the vast amounts of observation data, recursive methods are usually recommended. In this case, an iterative estimation of the requested quantities is performed, whereby the comprehensive total data set is divided into several individual epochs. If the most recent observations are successively available for each epoch, a filtering algorithm can be applied. Thus, an efficient estimation is carried out and, with respect to a comprehensive overall adjustment, generally larger observation sets can be considered. However, such filtering algorithms exist so far almost exclusively for explicit relations between the available observations and the requested estimation quantities. If this mathematical relationship is implicit, which is certainly the case for several practical issues, only a few methods exist or, in the case of recursive parameter estimation, none at all. This circumstance is accompanied by the fact that the combination of implicit relationships with constraints regarding the parameters to be estimated has not yet been investigated at all. In this thesis, a versatile filter algorithm is presented, which is valid for explicit and for implicit mathematical relations as well. For the first time, methods for the consideration of constraints are given, especially for implicit relations. The developed methodology will be comprehensively validated and evaluated by simulations and real-world application examples of practical relevance. The usage of real data is directly related to kinematic MSSs and the related tasks of calibration and georeferencing. The latter especially with regard to complex innercity environments. In such challenging environments, the requirements for georeferencing under integrity aspects are of special importance. Therefore, the simultaneous use of independent and complementary information sources is applied in this thesis. This enables a reliable georeferencing solution to be achieved and a prompt notification to be issued in case of integrity violations. Note de contenu : 1- Introduction 2- Fundamentals of Recursive State-space Filtering 3- Methodological contributions 4- Kinematic Multi-sensor Systems and Their Efficient Calibration 5- Information-based Georeferencing 6- Conclusions Numéro de notice : 17686 Affiliation des auteurs : non IGN Thématique : IMAGERIE/INFORMATIQUE Nature : Thèse étrangère Note de thèse : PhD thesis : Geodäsie und Geoinformatik : Hanovre : 2020 DOI : sans En ligne : https://dgk.badw.de/fileadmin/user_upload/Files/DGK/docs/c-856.pdf Format de la ressource électronique : URL Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=98164

Comparison of filtering algorithms used for DTM production from airborne lidar data: a case study in Bergama, Turkey / Baris Suleymanoglu in Geodetski vestnik, vol 63 n° 3 (September - November 2019)  

Public [article]  inGeodetski vestnik > vol 63 n° 3 (September - November 2019) . - pp 395 - 414 Titre : Comparison of filtering algorithms used for DTM production from airborne lidar data: a case study in Bergama, Turkey Type de document : Article/Communication Auteurs : Baris Suleymanoglu, Auteur ; Metin Soycan, Auteur Année de publication : 2019 Article en page(s) : pp 395 - 414 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Lasergrammétrie [Termes IGN] algorithme de filtrage [Termes IGN] analyse comparative [Termes IGN] convolution (signal) [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] modèle numérique de surface [Termes IGN] modèle numérique de terrain [Termes IGN] semis de points [Termes IGN] test de performance [Termes IGN] Turquie Résumé : (Auteur) A light detection and ranging (lidar) system is one of the most important technologies used for generating digital terrain models (DTMs). The point cloud data obtained by these systems consist of data gathered from ground and nonground features. To create a DTM with high resolution and accuracy, ground and nonground data must be separated. Numerous filtering algorithms have been developed for this purpose. The aim of this study was testing the filtering performance of six different filtering algorithms in four different test areas with different land cover were selected that had topographical features and characteristics. The algorithms were adaptive triangulated irregular network (ATIN), elevation threshold with an expand window (ETEW), maximum local slope (MLS), progressive morphology (PM), iterative polynomial fitting (IPF), and multiscale curvature classification (MCC) algorithms. In the results, all the filters performed well on a smooth surface and produced more errors in complex urban areas and rough terrain with dense vegetation. The IPF filtering algorithm generated the best results for the first three test areas (smooth landscape, urban areas and agricultural areas), while ETEW performed best in the fourth test area (steep areas with dense vegetation and infrastructure). Numéro de notice : A2019-502 Affiliation des auteurs : non IGN Thématique : POSITIONNEMENT Nature : Article DOI : 10.15292/geodetski-vestnik.2019.03.395-414 En ligne : http://dx.doi.org/10.15292/geodetski-vestnik.2019.03.395-414 Format de la ressource électronique : URL Article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=93784 [article]

Réservation

Réserver ce document

Exemplaires (1)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
139-2019031	SL	Revue	Centre de documentation	Revues en salle	Disponible

Recurrent neural networks to correct satellite image classification maps / Emmanuel Maggiori in IEEE Transactions on geoscience and remote sensing, vol 55 n° 9 (September 2017)  

Public [article]  inIEEE Transactions on geoscience and remote sensing > vol 55 n° 9 (September 2017) . - pp 4962 - 4971 Titre : Recurrent neural networks to correct satellite image classification maps Type de document : Article/Communication Auteurs : Emmanuel Maggiori, Auteur ; Guillaume Charpiat, Auteur ; Yuliya Tarabalka, Auteur ; Pierre Alliez, Auteur Année de publication : 2017 Article en page(s) : pp 4962 - 4971 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage automatique [Termes IGN] classification par réseau neuronal [Termes IGN] convolution (signal) [Termes IGN] itération [Termes IGN] réseau neuronal convolutif Résumé : (Auteur) While initially devised for image categorization, convolutional neural networks (CNNs) are being increasingly used for the pixelwise semantic labeling of images. However, the proper nature of the most common CNN architectures makes them good at recognizing but poor at localizing objects precisely. This problem is magnified in the context of aerial and satellite image labeling, where a spatially fine object outlining is of paramount importance. Different iterative enhancement algorithms have been presented in the literature to progressively improve the coarse CNN outputs, seeking to sharpen object boundaries around real image edges. However, one must carefully design, choose, and tune such algorithms. Instead, our goal is to directly learn the iterative process itself. For this, we formulate a generic iterative enhancement process inspired from partial differential equations, and observe that it can be expressed as a recurrent neural network (RNN). Consequently, we train such a network from manually labeled data for our enhancement task. In a series of experiments, we show that our RNN effectively learns an iterative process that significantly improves the quality of satellite image classification maps. Numéro de notice : A2017-659 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1109/TGRS.2017.2697453 En ligne : http://dx.doi.org/10.1109/TGRS.2017.2697453 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=87070 [article]

Analyse et traitement des signaux / E. Tisserand (2008)

Permalink

Detection and vectorization of roads from lidar data / S. Clode in Photogrammetric Engineering & Remote Sensing, PERS, vol 73 n° 5 (May 2007)

Permalink

A multi-resolution approach for filtering LiDAR altimetry data / José L. Silvan-Cardenas in ISPRS Journal of photogrammetry and remote sensing, vol 61 n° 1 (October 2006)  

Permalink

Hierarchical recovery of digital terrain models from single and multiple return lidar data / Y. Hu in Photogrammetric Engineering & Remote Sensing, PERS, vol 71 n° 4 (April 2005)

Permalink

Décomposition de signaux aléatoires stationnaires et non-stationnaires / Philippe Courmontagne (2005)

Permalink

Théorie élémentaire du signal / Luc Jolivet (2005)

Permalink

Analyse et traitement des signaux / E. Tisserand (2004)

Permalink

Cartographie des rizières d'une zone des hautes terres centrales de Madagascar pour la détermination des zones à risque du paludisme / F. Thomas (2004)

Permalink

Etude de méthodes d'inversion des données du Tams (télémètre laser aéroporté à champ large) pour la surveillance des mouvements verticaux du sol / Luc Mathis (2002)

Permalink

Modex / Stéphane Mallat (1998)

Permalink