Centre de documentation
scientifique

Single-image super-resolution for remote sensing images using a deep generative adversarial network with local and global attention mechanisms / Yadong Li in IEEE Transactions on geoscience and remote sensing, vol 60 n° 10 (October 2022)  

Public [article]  inIEEE Transactions on geoscience and remote sensing > vol 60 n° 10 (October 2022) . - n° 3000224 Titre : Single-image super-resolution for remote sensing images using a deep generative adversarial network with local and global attention mechanisms Type de document : Article/Communication Auteurs : Yadong Li, Auteur ; Sébastien Mavromatis, Auteur ; Feng Zhang, Auteur ; et al., Auteur Année de publication : 2022 Article en page(s) : n° 3000224 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage profond [Termes IGN] attention (apprentissage automatique) [Termes IGN] classification par réseau neuronal convolutif [Termes IGN] image isolée [Termes IGN] pouvoir de résolution géométrique [Termes IGN] pouvoir de résolution spectrale [Termes IGN] reconstruction d'image [Termes IGN] réseau antagoniste génératif Résumé : (auteur) Super-resolution (SR) technology is an important way to improve spatial resolution under the condition of sensor hardware limitations. With the development of deep learning (DL), some DL-based SR models have achieved state-of-the-art performance, especially the convolutional neural network (CNN). However, considering that remote sensing images usually contain a variety of ground scenes and objects with different scales, orientations, and spectral characteristics, previous works usually treat important and unnecessary features equally or only apply different weights in the local receptive field, which ignores long-range dependencies; it is still a challenging task to exploit features on different levels and reconstruct images with realistic details. To address these problems, an attention-based generative adversarial network (SRAGAN) is proposed in this article, which applies both local and global attention mechanisms. Specifically, we apply local attention in the SR model to focus on structural components of the earth’s surface that require more attention, and global attention is used to capture long-range interdependencies in the channel and spatial dimensions to further refine details. To optimize the adversarial learning process, we also use local and global attentions in the discriminator model to enhance the discriminative ability and apply the gradient penalty in the form of hinge loss and loss function that combines L1 pixel loss, L1 perceptual loss, and relativistic adversarial loss to promote rich details. The experiments show that SRAGAN can achieve performance improvements and reconstruct better details compared with current state-of-the-art SR methods. A series of ablation investigations and model analyses validate the efficiency and effectiveness of our method. Numéro de notice : A2022-767 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1109/TGRS.2021.3093043 Date de publication en ligne : 12/07/2021 En ligne : https://doi.org/10.1109/TGRS.2021.3093043 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=101789 [article]

Encoder-decoder structure with multiscale receptive field block for unsupervised depth estimation from monocular video / Songnan Chen in Remote sensing, Vol 14 n° 12 (June-2 2022)  

Public [article]  inRemote sensing > Vol 14 n° 12 (June-2 2022) . - n° 2906 Titre : Encoder-decoder structure with multiscale receptive field block for unsupervised depth estimation from monocular video Type de document : Article/Communication Auteurs : Songnan Chen, Auteur ; Junyu Han, Auteur ; Mengxia Tang, Auteur ; et al., Auteur Année de publication : 2022 Article en page(s) : n° 2906 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage non-dirigé [Termes IGN] classification par réseau neuronal convolutif [Termes IGN] couple stéréoscopique [Termes IGN] données d'entrainement (apprentissage automatique) [Termes IGN] image isolée [Termes IGN] optimisation (mathématiques) [Termes IGN] profondeur [Termes IGN] séquence d'images [Termes IGN] structure-from-motion Résumé : (auteur) Monocular depth estimation is a fundamental yet challenging task in computer vision as depth information will be lost when 3D scenes are mapped to 2D images. Although deep learning-based methods have led to considerable improvements for this task in a single image, most existing approaches still fail to overcome this limitation. Supervised learning methods model depth estimation as a regression problem and, as a result, require large amounts of ground truth depth data for training in actual scenarios. Unsupervised learning methods treat depth estimation as the synthesis of a new disparity map, which means that rectified stereo image pairs need to be used as the training dataset. Aiming to solve such problem, we present an encoder-decoder based framework, which infers depth maps from monocular video snippets in an unsupervised manner. First, we design an unsupervised learning scheme for the monocular depth estimation task based on the basic principles of structure from motion (SfM) and it only uses adjacent video clips rather than paired training data as supervision. Second, our method predicts two confidence masks to improve the robustness of the depth estimation model to avoid the occlusion problem. Finally, we leverage the largest scale and minimum depth loss instead of the multiscale and average loss to improve the accuracy of depth estimation. The experimental results on the benchmark KITTI dataset for depth estimation show that our method outperforms competing unsupervised methods. Numéro de notice : A2022-563 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.3390/rs14122906 En ligne : https://doi.org/10.3390/rs14122906 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=101240 [article]

Building footprint extraction in Yangon city from monocular optical satellite image using deep learning / Hein Thura Aung in Geocarto international, vol 37 n° 3 ([01/02/2022])  

Public [article]  inGeocarto international > vol 37 n° 3 [01/02/2022] . - pp 792 - 812 Titre : Building footprint extraction in Yangon city from monocular optical satellite image using deep learning Type de document : Article/Communication Auteurs : Hein Thura Aung, Auteur ; Sao Hone Pha, Auteur ; Wataru Takeuchi, Auteur Année de publication : 2022 Article en page(s) : pp 792 - 812 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage profond [Termes IGN] Birmanie [Termes IGN] détection du bâti [Termes IGN] empreinte [Termes IGN] image Geoeye [Termes IGN] image isolée [Termes IGN] réseau antagoniste génératif [Termes IGN] vision monoculaire Résumé : (auteur) In this research, building footprints in Yangon City, Myanmar are extracted only from monocular optical satellite image by using conditional generative adversarial network (CGAN). Both training dataset and validating dataset are created from GeoEYE image of Dagon Township in Yangon City. Eight training models are created according to the change of values in three training parameters; learning rate, β1 term of Adam, and number of filters in the first convolution layer of the generator and the discriminator. The images of the validating dataset are divided into four image groups; trees, buildings, mixed trees and buildings, and pagodas. The output images of eight trained models are transformed to the vector images and then evaluated by comparing with manually digitized polygons using completeness, correctness and F1 measure. According to the results, by using CGAN, building footprints can be extracted up to 71% of completeness, 81% of correctness and 69% of F1 score from only monocular optical satellite image. Numéro de notice : A2022-345 Affiliation des auteurs : non IGN Thématique : IMAGERIE/INFORMATIQUE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1080/10106049.2020.1740949 Date de publication en ligne : 20/03/2020 En ligne : https://doi.org/10.1080/10106049.2020.1740949 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100526 [article]

Monocular depth estimation in forest environments / Hristina Hristova (2022)  

Public Titre : Monocular depth estimation in forest environments Type de document : Article/Communication Auteurs : Hristina Hristova, Auteur ; Meinrad Abegg, Auteur ; Christoph Fischer, Auteur ; Nataliia Rehush, Auteur Editeur : International Society for Photogrammetry and Remote Sensing ISPRS Année de publication : 2022 Collection : International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, ISSN 1682-1750 num. 43-B2 Conférence : ISPRS 2022, Commission 2, 24th ISPRS international congress, Imaging today, foreseeing tomorrow 06/06/2022 11/06/2022 Nice France OA ISPRS Archives Importance : pp 1017 - 1023 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Lasergrammétrie [Termes IGN] apprentissage dirigé [Termes IGN] données lidar [Termes IGN] données localisées 3D [Termes IGN] forêt [Termes IGN] image isolée [Termes IGN] inventaire forestier (techniques et méthodes) [Termes IGN] jeu de données localisées [Termes IGN] profondeur [Termes IGN] vision monoculaire Résumé : (auteur) Depth estimation from a single image is a challenging task, especially inside the highly structured forest environment. In this paper, we propose a supervised deep learning model for monocular depth estimation based on forest imagery. We train our model on a new data set of forest RGB-D images that we collected using a terrestrial laser scanner. Alongside the input RGB image, our model uses a sparse depth channel as input to recover the dense depth information. The prediction accuracy of our model is significantly higher than that of state-of-the-art methods when applied in the context of forest depth estimation. Our model brings the RMSE down to 2.1 m, compared to 4 m and above for reference methods. Numéro de notice : C2022-022 Affiliation des auteurs : non IGN Thématique : FORET/IMAGERIE/INFORMATIQUE Nature : Communication nature-HAL : ComAvecCL&ActesPubliésIntl DOI : 10.5194/isprs-archives-XLIII-B2-2022-1017-2022 Date de publication en ligne : 30/05/2022 En ligne : http://dx.doi.org/10.5194/isprs-archives-XLIII-B2-2022-1017-2022 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=100848

Single-image photogrammetry for deriving tree architectural traits in mature forest stands: a comparison with terrestrial laser scanning / Kamil Kędra in Annals of Forest Science, vol 76 n° 1 (March 2019)  

Public [article]  inAnnals of Forest Science > vol 76 n° 1 (March 2019) Titre : Single-image photogrammetry for deriving tree architectural traits in mature forest stands: a comparison with terrestrial laser scanning Type de document : Article/Communication Auteurs : Kamil Kędra, Auteur ; Ignacio Barbeito, Auteur ; Mathieu Dassot , Auteur ; Patrick Vallet, Auteur ; Anna Gazda, Auteur Année de publication : 2019 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications photogrammétriques [Termes IGN] allométrie [Termes IGN] analyse comparative [Termes IGN] coefficient de corrélation [Termes IGN] détection d'arbres [Termes IGN] diamètre à hauteur de poitrine [Termes IGN] diamètre des arbres [Termes IGN] forêt tempérée [Termes IGN] hauteur des arbres [Termes IGN] houppier [Termes IGN] image isolée [Termes IGN] Orléans, forêt domaniale d' (Loiret) [Termes IGN] peuplement forestier [Termes IGN] peuplement mélangé [Termes IGN] Pinus sylvestris [Termes IGN] Quercus (genre) [Termes IGN] télémétrie laser terrestre Résumé : (auteur) Key message : We compared two methods for detailed individual tree measurements: single image photogrammetry (SIP), a simplified, low-cost method, and the state-of-the-art terrestrial laser scanning (TLS). Our results provide evidence that SIP can be successfully applied to obtain accurate tree architectural traits in mature forests. Context : Tree crown variables are necessary in forest modelling; however, they are time consuming to measure directly, and they are measured in many different ways. We compare two methods to obtain crown variables: laser-based and image-based. TLS is an advanced technology for three-dimensional data acquisition; SIP is a simplified, low-cost method. Aims : To elucidate differences between the methods, and validate SIP accuracy and usefulness for forest research, we investigated if (1) SIP and TLS measurements are in agreement in terms of the most widely used tree characteristics; (2) differences between the SIP traits and their TLS counterparts are constant throughout tree density and species composition; (3) tree architectural traits obtained with SIP explain differences in laser-based crown projection area (CPA), under different forest densities and stand compositions; and (4) CPA modelled with SIP variables is more accurate than CPA obtained with stem diameter-based allometric models. We also examined the correspondence between local tree densities extracted from images and from field measurements. Methods : We compared TLS and SIP in a temperate pure sessile oak and mixed with Scots pine stands, in the Orléans Forest, France. Standard major axis regression was used to establish relations between laser-based and image-based tree height and diameter at breast height. Four SIP-derived traits were compared between the levels of stand density and species composition with a t test, in terms of deviations and biases to their TLS counterparts. We created a set of linear and linear mixed models (LMMs) of CPATLS, with SIP variables. Both laser-based and image-based stem diameters were used to estimate CPA with the published allometric equations; the results were then compared with the best predictive LMM, in terms of similarity with CPATLS measurement. Local tree density extracted from images was compared with field measurements in terms of basic statistics and correlation. Results : Tree height and diameter at breast height were reliably represented by SIP (Pearson correlation coefficients r = 0.92 and 0.97, respectively). SIP measurements were affected by the stand composition factor; tree height attained higher mean absolute deviation (1.09 m) in mixed stands, compared to TLS, than in pure stands (0.66 m); crown width was more negatively biased in mixed stands (− 0.79 m), than in pure stands (− 0.05 m); and diameter at breast height and crown asymmetry were found unaffected. Crown width and mean branch angle were key SIP explanatory variables to predict CPATLS. The model was approximately 2-fold more accurate than the CPA allometric estimations with both laser-based and image-based stem diameters. SIP-derived local tree density was similar to the field-measured density in terms of mean and standard deviation (9.6 (3.5) and 9.4 (3.6) trees per plot, respectively); the correlation between both density measures was significantly positive (r = 0.76). Conclusion : SIP-derived variables, such as crown width, mean branch angle, branch thickness, and crown asymmetry, were useful to explain tree architectural differences under different densities and stand compositions and may be implemented in many forest research applications. SIP may also provide a coarse measure of local competition, in terms of number of neighbouring trees. Our study provides the first test in mature forest stands, for SIP compared with TLS. Numéro de notice : A2019-044 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1007/s13595-018-0783-x Date de publication en ligne : 07/01/2019 En ligne : https://doi.org/10.1007/s13595-018-0783-x Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=92050 [article]

Complete 3D scene parsing from an RGBD image / Chuhang Zou in International journal of computer vision, vol 127 n° 2 (February 2019)  

Permalink

Single image dehazing via an improved atmospheric scattering model / Mingye Ju in The Visual Computer, vol 33 n° 12 (December 2017)  

Permalink

Probabilistic multi-person localisation and tracking / Tobias Klinger (2016) 

Permalink

Object-based building change detection from a single multispectral image and pre-existing geospatial information / Georgia Doxani in Photogrammetric Engineering & Remote Sensing, PERS, vol 81 n° 6 (June 2015)  

Permalink

Nouvelles cartographies au 1:200 000 du Sénégal et au 1:5 000 en Polynésie Française : deux exemples de production opérationnelle de cartes topographiques numériques à partir d'imagerie satellitaire / Dominique Lasselin in Le monde des cartes, n° 202 (décembre 2009) 

Permalink

A combined single range and single image device for low-cost measurement of building façade features / C. Ordonez in Photogrammetric record, vol 23 n° 122 (June - August 2008)  

Permalink

From 2D to 3D through modelling based on a single image / Diego Gonzalez-Aguilera in Photogrammetric record, vol 23 n° 122 (June - August 2008)  

Permalink

Détection automatique des points de fuite et calcul de leur incertitude à l'aide de la géométrie projective / Mahzad Kalantari (2008)    

Permalink

Robust and automatic vanishing points detection with their uncertainties from a single uncalibrated image, by planes extraction on the unit sphere / Mahzad Kalantari (2008)    

Permalink

A stochastic framework for the identification of building rooftops using a single remote sensing image / A. Katartzis in IEEE Transactions on geoscience and remote sensing, vol 46 n° 1 (January 2008)  

Permalink