Centre de documentation
scientifique

A comparison of fuzzy vs. augmented-ISODATA classification algorithms for cloud-shadow discrimination from Landsat images / A.M. Melesse in Photogrammetric Engineering & Remote Sensing, PERS, vol 68 n° 9 (September 2002)  

Public [article]  inPhotogrammetric Engineering & Remote Sensing, PERS > vol 68 n° 9 (September 2002) . - pp 905 - 911 Titre : A comparison of fuzzy vs. augmented-ISODATA classification algorithms for cloud-shadow discrimination from Landsat images Type de document : Article/Communication Auteurs : A.M. Melesse, Auteur ; J.D. Jordan, Auteur Année de publication : 2002 Article en page(s) : pp 905 - 911 Note générale : Bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Traitement d'image optique [Termes IGN] analyse comparative [Termes IGN] bande visible [Termes IGN] classification dirigée [Termes IGN] classification floue [Termes IGN] classification ISODATA [Termes IGN] image Landsat-TM [Termes IGN] nuage [Termes IGN] ombre Résumé : (Auteur) Satellite images are the most important source of landcover data over a large range of temporal and spatial scales. However, the complete realization of satellite imagery as a source of landcover information is limited by the presence of contaminants such as cloud and associated shadows in the image. These contaminants are not adequately handled with conventional image classification techniques such as the unsupervised maximumlikelihood technique. This study comprises a comparison of two classification algorithms, the fuzzy technique and an augmented form of the Iterative SelfOrganizing Data Analysis (ISODATA) technique, which were used to discriminate lowaltitude clouds and their shadows on a Landsat Thematic Mapper (TM) image of the Econlockhatchee River basin (Econ), in central Florida. Preliminary conventional unsupervised maxim umlikelihood classification of the image resulted in clouds being mixed with builtups and shadows being mixed with water bodies. Regions containing these two kinds of mixed categories were first masked, then fuzzy and augmented ISODATA classifications were performed on them. The ISODATA classification algorithm was run on the TM visible/shortwave bands and augmented with scatter diagrams of surface temperature versus several vegetation indices; the fuzzy algorithm was run on TM bands 1 through 5 and band 7. An accuracy assessment of the techniques was carried out using 40 randomly selected points within the image. Results of the classifications showed that both algorithms successfully discriminated clouds from other bright features, and shadows from other dark features, with an overall accuracy of greater than 80 percent. Numéro de notice : A2002-182 Affiliation des auteurs : non IGN Thématique : IMAGERIE Nature : Article DOI : sans En ligne : https://www.asprs.org/wp-content/uploads/pers/2002journal/september/2002_sep_905 [...] Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=22097 [article]