ML estimation of wavelet regularization hyperparameters in inverse problems.
Roberto CavicchioliCaroline ChauxLaure Blanc-FéraudLuca ZanniPublished in: ICASSP (2013)
Keyphrases
- inverse problems
- hyperparameters
- maximum likelihood
- image reconstruction
- maximum a posteriori
- global optimization
- em algorithm
- regularization parameter
- model selection
- cross validation
- closed form
- convex optimization
- bayesian inference
- bayesian framework
- random sampling
- optimization methods
- parameter space
- prior information
- early vision
- support vector
- incremental learning
- smoothness constraint
- optimization problems
- denoising
- noise level
- partial differential equations
- incomplete data
- pairwise
- multiscale
- high quality
- wavelet transform
- sample size
- active learning
- expectation maximization
- image segmentation
- multiresolution
- prior knowledge
- machine learning
- wavelet coefficients
- super resolution
- level set