Analysis of Land Cover Type Using Landsat-8 Data - Inria - Institut national de recherche en sciences et technologies du numérique Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Analysis of Land Cover Type Using Landsat-8 Data

Résumé

Classification of images attributes to categorizing of images into various predefined groups. A particular image can be grouped into several diverse classes. Examining and ordering the images manually is a tiresome job particularly when they are abundant and therefore, automating the entire process using image processing and computer vision would be very efficient and useful. In this study, the Classifier and Regression trees (CART) algorithm is used to create a classifier model that classifies a region based on the feature specified. The Google Earth Engine (GEE) platform is utilized to conduct the study. The Tier 1 USGS Landsat 8 surface reflectance dataset is employed and is sorted according to the cloud cover. The features are then extracted and are merged to obtain a feature collection. This input imagery is further sampled using particular bands from the Landsat imagery to get a renewed feature collection of training data and the classifier model is trained using the CART Algorithm. An accuracy assessment is further performed to determine the exactness of the proposed model and the results are plotted using a confusion matrix. By applying the CART algorithm for image classification, an accuracy of 83% is achieved which was found to be better than the existing results.
Fichier principal
Vignette du fichier
512058_1_En_8_Chapter.pdf (15.45 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03772946 , version 1 (08-09-2022)

Licence

Paternité

Identifiants

Citer

V. Samuktha, M. Sabeshnav, A. Krishna Sameera, J. Aravinth, S. Veni. Analysis of Land Cover Type Using Landsat-8 Data. 4th International Conference on Computational Intelligence in Data Science (ICCIDS), Mar 2021, Chennai, India. pp.78-89, ⟨10.1007/978-3-030-92600-7_8⟩. ⟨hal-03772946⟩
38 Consultations
3 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More