Using Bayesian optimization to jointly tune the classifier and the random field for spatial-spectral hyperspectral classification.
Utsav B. GewaliSildomar T. MonteiroPublished in: IGARSS (2017)
Keyphrases
- hyperspectral
- pixel classification
- hyperspectral data
- hyperspectral images
- random fields
- hyperspectral imagery
- band selection
- hyperspectral image classification
- spectral signatures
- remote sensing
- multispectral
- infrared
- hyperspectral remote sensing
- spectral imagery
- spectral bands
- image data
- hyperspectral imaging
- target detection
- support vector
- non stationary
- spatial resolution
- support vector machine
- markov random field
- feature selection
- maximum entropy
- information content
- bayesian classifier
- svm classifier
- feature space
- spectral data
- pattern recognition
- reflectance spectra
- bayesian networks
- high quality
- feature set
- image segmentation
- spectral resolution
- pairwise
- spatial information
- satellite images
- training data
- active learning
- probabilistic model
- parameter estimation
- computer vision
- machine learning