Centre dedocumentation
scientifique

Dépouillements

Band grouping versus band clustering in SVM ensemble classification of hyperspectral imagery / Behnaz Bigdeli in Photogrammetric Engineering & Remote Sensing, PERS, vol 79 n° 6 (June 2013)  

Public [article]  inPhotogrammetric Engineering & Remote Sensing, PERS > vol 79 n° 6 (June 2013) . - pp 523 - 533 Titre : Band grouping versus band clustering in SVM ensemble classification of hyperspectral imagery Type de document : Article/Communication Auteurs : Behnaz Bigdeli, Auteur ; Farhad Samadzadegan, Auteur ; Peter Reinartz, Auteur Année de publication : 2013 Article en page(s) : pp 523 - 533 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] classification par séparateurs à vaste marge [Termes IGN] image hyperspectrale [Termes IGN] regroupement de données Résumé : (Auteur) Due to the dense sampling of spectral signatures of land covers, hyperspectral images have a better discrimination among similar ground cover classes than traditional remote sensing data. However, these images are usually composed of tens or hundreds of spectrally close bands, which result in high redundancy and great amount of computation time in hyperspectral image classification. In addition, the large number of spectral bands, but limited availability of training samples creates the problem of Hughes phenomenon. Consequently, traditional classification strategies have often limited performance in classification of hyperspectral imagery. Referring to the limitation of single classifiers in these situations, classifier ensemble system may exhibit better performance. This paper presents a method for classification of hyperspectral data based on two concepts of Band Clustering (BC) and Band Grouping [eg] through a Support Vector machine (SVM) ensemble system. The proposed method uses the BC\BG strategies to split data into few band portions. After this step, we applied SVM on each band cluster\group that is produced in previous step. Finally, Naive Bayes as a classifier fusion method combines the decisions of SVM classifiers. Experimental results show that the proposed method improves the classification accuracy in comparison to the standard SVM and to feature selection methods. Numéro de notice : A2013-362 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.14358/PERS.79.6.523 En ligne : https://doi.org/10.14358/PERS.79.6.523 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=32500 [article]

An automated approach for the conflation of vector parcel map with imagery / Wenbo Song in Photogrammetric Engineering & Remote Sensing, PERS, vol 79 n° 6 (June 2013)  

Public [article]  inPhotogrammetric Engineering & Remote Sensing, PERS > vol 79 n° 6 (June 2013) . - pp 535 - 543 Titre : An automated approach for the conflation of vector parcel map with imagery Type de document : Article/Communication Auteurs : Wenbo Song, Auteur ; James M. Keller, Auteur ; Timothy L. Haithcoat, Auteur ; et al., Auteur Année de publication : 2013 Article en page(s) : pp 535 - 543 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image [Termes IGN] appariement d'images [Termes IGN] cohérence géométrique [Termes IGN] conflation [Termes IGN] données vectorielles [Termes IGN] image à haute résolution [Termes IGN] plan parcellaire [Termes IGN] précision du positionnement Résumé : (Auteur) Local governments frequently use parcel maps or other spatial data for decision-making, but much of this data is often inaccurate and outdated, producing a negative influence on the expected outcome. Remote sensing can provide accurate and current high-resolution imagery. However, one major bottleneck to the integration of high-resolution imagery into existing geographic information systems is the issue of positional accuracy of the existing line-work within the vector GIS database, making it difficult to match with imagery. This paper presents a vector-to-imagery conflation approach to improve the positional accuracy of digital parcel maps by conflating the vector parcel maps to make it consistent with high-resolution imagery. Road intersections are automatically extracted from imagery and used as control points. A relaxation labeling algorithm is used to find the matches between the two road intersection point sets from vector and imagery. The links are created from the matched point pairs and are used to perform a piecewise transformation. The test results show that this approach can improve the accuracy of vector parcel maps significantly. It is a very cost-effective method and has great potential to save considerable time and money for local governments to upgrade their inaccurate vector datasets. Numéro de notice : A2013-363 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : 10.14358/PERS.79.6.535 En ligne : https://doi.org/10.14358/PERS.79.6.535 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=32501 [article]