Object-based random forest classification of Landsat ETM+ and WorldView-2 satellite imagery for mapping lowland native grassland communities in Tasmania, Australia.
Bethany MelvilleArko LucieerJagannath AryalPublished in: Int. J. Appl. Earth Obs. Geoinformation (2018)
Keyphrases
- random forest
- satellite imagery
- decision trees
- fold cross validation
- feature set
- landsat etm
- remote sensing images
- random forests
- decision tree learning algorithms
- change detection
- pattern recognition
- feature vectors
- urban areas
- classification accuracy
- satellite images
- classification models
- ensemble methods
- ensemble classifier
- support vector
- machine learning
- remote sensing
- machine learning methods
- image classification
- training samples
- support vector machine svm
- multispectral
- feature space
- feature extraction
- supervised learning
- land cover
- class labels
- base classifiers
- unsupervised learning
- simulated annealing
- support vector machine
- evolutionary algorithm
- feature selection