Sparse representation of hyperspectral data using CUR matrix decomposition.
Jakob SigurdssonMagnus O. UlfarssonJohannes R. SveinssonJón Atli BenediktssonPublished in: IGARSS (2013)
Keyphrases
- matrix decomposition
- sparse representation
- random projections
- nonnegative matrix factorization
- low rank
- negative matrix factorization
- low rank approximation
- high dimensional data
- subspace learning
- reconstruction error
- sparse coding
- low rank matrix
- singular values
- face recognition
- singular value decomposition
- missing data
- image classification
- dimensionality reduction
- signal processing
- kernel matrix
- matrix factorization
- convex optimization
- linear combination
- lower dimensional
- image representation
- principal component analysis
- data matrix
- original data
- high dimensionality
- image processing
- low dimensional
- data sets
- high order
- similarity search
- least squares
- nearest neighbor
- high dimensional
- image segmentation