Centre de documentation
scientifique

Multi-view urban scene classification with a complementary-information learning model / Wanxuan Geng in Photogrammetric Engineering & Remote Sensing, PERS, vol 88 n° 1 (January 2022)  

Public [article]  inPhotogrammetric Engineering & Remote Sensing, PERS > vol 88 n° 1 (January 2022) . - pp 65 - 72 Titre : Multi-view urban scene classification with a complementary-information learning model Type de document : Article/Communication Auteurs : Wanxuan Geng, Auteur ; Weixun Zhou, Auteur ; Shuanggen Jin, Auteur Année de publication : 2022 Article en page(s) : pp 65 - 72 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] apprentissage automatique [Termes IGN] classification par séparateurs à vaste marge [Termes IGN] données de terrain [Termes IGN] données multisources [Termes IGN] extraction de traits caractéristiques [Termes IGN] fusion de données multisource [Termes IGN] image aérienne [Termes IGN] niveau du sol [Termes IGN] précision de la classification [Termes IGN] scène urbaine Résumé : (Auteur) Traditional urban scene-classification approaches focus on images taken either by satellite or in aerial view. Although single-view images are able to achieve satisfactory results for scene classification in most situations, the complementary information provided by other image views is needed to further improve performance. Therefore, we present a complementary information-learning model (CILM) to perform multi-view scene classification of aerial and ground-level images. Specifically, the proposed CILM takes aerial and ground-level image pairs as input to learn view-specific features for later fusion to integrate the complementary information. To train CILM, a unified loss consisting of cross entropy and contrastive losses is exploited to force the network to be more robust. Once CILM is trained, the features of each view are extracted via the two proposed feature-extraction scenarios and then fused to train the support vector machine classifier for classification. The experimental results on two publicly available benchmark data sets demonstrate that CILM achieves remarkable performance, indicating that it is an effective model for learning complementary information and thus improving urban scene classification. Numéro de notice : A2022-063 Affiliation des auteurs : non IGN Thématique : IMAGERIE/URBANISME Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.14358/PERS.21-00062R2 Date de publication en ligne : 01/01/2022 En ligne : https://doi.org/10.14358/PERS.21-00062R2 Format de la ressource électronique : URL Article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=99708 [article]

Exemplaires(1)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
105-2022011	SL	Revue	Centre de documentation	Revues en salle	Disponible

Phenomenology of ground scattering in a tropical forest through polarimetric synthetic aperture radar tomography / Mauro Mariotti d'Alessandro in IEEE Transactions on geoscience and remote sensing, vol 51 n° 8 (August 2013)  

Public [article]  inIEEE Transactions on geoscience and remote sensing > vol 51 n° 8 (August 2013) . - pp 4430 - 4437 Titre : Phenomenology of ground scattering in a tropical forest through polarimetric synthetic aperture radar tomography Type de document : Article/Communication Auteurs : Mauro Mariotti d'Alessandro, Auteur ; Stefano Tebaldini, Auteur ; Fabio Rocca, Auteur Année de publication : 2013 Article en page(s) : pp 4430 - 4437 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image radar et applications [Termes IGN] bande P [Termes IGN] forêt tropicale [Termes IGN] Guyane (département français) [Termes IGN] image radar moirée [Termes IGN] niveau du sol [Termes IGN] polarimétrie radar [Termes IGN] tomographie Résumé : (Auteur) This paper aims at characterizing the scattering mechanisms occurring at the ground level in a tropical forest illuminated by a P-band synthetic aperture radar (SAR). The analysis is carried out based on the multibaseline, fully polarimetric, data set collected by ONERA over Paracou, French Guyana, in the frame of the European space agency campaign TropiSAR. The favorable baseline distribution of this data set results in the possibility of removing most contributions from the vegetation layer by tomographic techniques, thus allowing the generation of a new fully polarimetric single look complex SAR image relative to scattering contributions from the ground level only. Such a ground layer image is then analyzed by considering the variation of its polarimetric signature with respect to terrain local slope and Radar look angle. Two major conclusions are drawn: 1) double bounce scattering from trunk-ground interactions is observed to be the dominant scattering mechanism at the ground level on flat terrains, whereas it rapidly tends to vanish as the topographic slope increases, and 2) the characteristic parameter that rules trunk-ground scattering is not the tree height, but rather the available free path facing the tree, as a result of the presence of nearby trees, undulating topography, or understory preventing double bounce scattering from taking place whenever the ground bounce occurs too far away from the considered tree. The mean free path length resulting from the analysis of this data-set is found to be L ? 7 m. Finally, we discuss how the concept of free path length can be accounted for in simple terms by assuming an equivalent extinction model characterized by a variation along the horizontal dimension. Numéro de notice : A2013-415 Affiliation des auteurs : non IGN Thématique : FORET/IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.1109/TGRS.2013.2246573 En ligne : https://doi.org/10.1109/TGRS.2013.2246573 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=32553 [article]

Exemplaires(1)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
065-2013081	RAB	Revue	Centre de documentation	En réserve L003	Disponible

Robust reconstruction of building models from three-dimensional line segments / Jiann-Yeou Rau in Photogrammetric Engineering & Remote Sensing, PERS, vol 69 n° 2 (February /2003)  

Public [article]  inPhotogrammetric Engineering & Remote Sensing, PERS > vol 69 n° 2 (February /2003) . - pp 181 - 188 Titre : Robust reconstruction of building models from three-dimensional line segments Type de document : Article/Communication Auteurs : Jiann-Yeou Rau, Auteur ; L.C. Chen, Auteur Année de publication : 2003 Article en page(s) : pp 181 - 188 Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Photogrammétrie spatiale [Termes IGN] automatisation [Termes IGN] contour [Termes IGN] extraction semi-automatique [Termes IGN] méthode robuste [Termes IGN] niveau du sol [Termes IGN] primitive volumique [Termes IGN] reconstruction 3D du bâti [Termes IGN] segment de droite Résumé : (Auteur) This paper presents a novel method for semiautomatically constructing building models from photogrammetric 3D line segments of buildings, i.e., their roof edges. The method, which we call "Split MergeShape" (SMS), can treat both complete line segments as well as incomplete line segments due to image occlusions. The proposed method is comprised of five major parts: (1) the creation of the Region of Interest (ROI) and preprocessing, (2) splitting the model by using the 3D line segments to construct a combination of roof primitives, (3) merging connected roof primitives to complete the boundary of each building, (4) shaping each building rooftop by connected coplanar analysis and coplanar fitting, and (5) quality assurance. The experimental results indicate that the proposed method can soundly rebuild the topology from the 3D line segments and reconstruct building models with up to a 98 percent success rate. The proposed SMS method has been proved reliable and effective, with a high degree of automation, even when groups of connected buildings or complex types of buildings are processed. Numéro de notice : A2003-013 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.14358/PERS.69.2.181 En ligne : https://doi.org/10.14358/PERS.69.2.181 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=22311 [article]