Centre de documentation
scientifique

Individual tree detection and classification for mapping pine wilt disease using multispectral and visible color imagery acquired from unmanned aerial vehicle / Takeshi Hoshikawa in Journal of The Remote Sensing Society of Japan, vol 40 n° 1 (2020)  

Public [article]  inJournal of The Remote Sensing Society of Japan > vol 40 n° 1 (2020) . - pp 13 - 19 Titre : Individual tree detection and classification for mapping pine wilt disease using multispectral and visible color imagery acquired from unmanned aerial vehicle Type de document : Article/Communication Auteurs : Takeshi Hoshikawa, Auteur ; Kazukiyo Yamamoto, Auteur Année de publication : 2020 Article en page(s) : pp 13 - 19 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications de télédétection [Termes IGN] carte de la végétation [Termes IGN] détection d'arbres [Termes IGN] image captée par drone [Termes IGN] image multibande [Termes IGN] indice de végétation [Termes IGN] maladie phytosanitaire [Termes IGN] modèle de régression [Termes IGN] Normalized Difference Vegetation Index [Termes IGN] Pinus (genre) [Termes IGN] protection des forêts [Termes IGN] régression logistique [Termes IGN] semis de points Résumé : (auteur) Pine wilt disease is one of the most destructive disease of pine forests. It is important to detect and exterminate infected trees for preservation of the forest. We demonstrated a novel method combining individual tree detection (ITD) and classification by logistic regression using unmanned aerial vehicle (UAV) images for the mapping of infected trees. In the ITD phase, 50 % and 84 % of damaged trees were automatically detected from the 3D point cloud generated from the UAV images using the local maximum filter. These rates of detection were comparable to previous studies that used UAV imagery. Subsequently, five vegetation indices calculated from multispectral and visible color (RGB) images were used. Among the vegetation indices, normalized difference vegetation index (NDVI), normalized difference red edge index (NDRE), and vegetation atmospherically resistant index (VARI) were preferable explanatory variable in the logistic regression to divide damaged and undamaged trees. The accuracy of these models ranged from 98 % to 100 % and the F-measure ranged from 94 % to 100 %. The best model, the logistic regression model using VARI as the explanatory variable, was then tested using five datasets to evaluate general performance. Each model showed explicitly high accuracy ranging from 95 % to 100 %. The best accuracy when considering the ITD and classification was 84 %. To map pine wilt disease, the proposed method is suitable for practical use due to its high-efficient and low-cost. Numéro de notice : A2020-405 Affiliation des auteurs : non IGN Thématique : FORET/IMAGERIE Nature : Article DOI : 10.11440/rssj.40.13 Date de publication en ligne : 31/01/2020 En ligne : https://doi.org/10.11440/rssj.40.13 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=96090 [article]