A fundamental pitfall in blind deconvolution with sparse and shift-invariant priors.
Alexis BenichouxEmmanuel VincentRémi GribonvalPublished in: ICASSP (2013)
Keyphrases
- shift invariant
- blind deconvolution
- blurred images
- sparse coding
- image restoration
- bayesian framework
- synthetic and real images
- point spread function
- map estimation
- image prior
- spatial domain
- motion blur
- wavelet domain
- image deblurring
- total variation
- sparse representation
- wavelet coefficients
- generative model
- natural images
- maximum a posteriori
- computer vision
- edge information
- edge preserving
- image structure
- prior knowledge
- linear combination
- image segmentation
- image quality
- multi frame
- markov random field
- wavelet transform
- pixel values
- multiresolution
- noise reduction