A novel non-parametric weighted feature extraction method for classification of hyperspectral image with limited training samples.
Jinn-Min YangBor-Chen KuoHsiao-Yun HuangPao-Ta YuPublished in: IGARSS (2007)
Keyphrases
- training samples
- hyperspectral images
- feature space
- training set
- learning algorithm
- supervised learning
- decision boundary
- training data
- test sample
- hyperplane
- number of training samples
- high dimensional
- nearest neighbor classifier
- hyperspectral
- remote sensing
- hyperspectral imagery
- hyperspectral data
- support vector machine
- data sets
- representative samples
- computer vision
- hyperspectral image classification
- kernel function
- active learning
- decision trees
- machine learning