Matrix computations (3rd ed.)
Images as Embedded Maps and Minimal Surfaces: Movies, Color, Texture, and Volumetric Medical Images
International Journal of Computer Vision - Special issue on computer vision research at the Technion
Digital Image Processing
Gabor Feature Space Diffusion via the Minimal Weighted Area Method
EMMCVPR '01 Proceedings of the Third International Workshop on Energy Minimization Methods in Computer Vision and Pattern Recognition
Image Deblurring in the Presence of Impulsive Noise
International Journal of Computer Vision
The edge preserving wiener filter for scalar and tensor valued images
DAGM'06 Proceedings of the 28th conference on Pattern Recognition
Variational deblurring of images with uncertain and spatially variant blurs
PR'05 Proceedings of the 27th DAGM conference on Pattern Recognition
-SVD: An Algorithm for Designing Overcomplete Dictionaries for Sparse Representation
IEEE Transactions on Signal Processing
Deterministic edge-preserving regularization in computed imaging
IEEE Transactions on Image Processing
Color TV: total variation methods for restoration of vector-valued images
IEEE Transactions on Image Processing
A general framework for low level vision
IEEE Transactions on Image Processing
An adaptive Gaussian model for satellite image deblurring
IEEE Transactions on Image Processing
Semi-blind image restoration via Mumford-Shah regularization
IEEE Transactions on Image Processing
Hi-index | 0.00 |
Most inverse problems require a regularization term on the data. The classic approach for the variational formulation is to use the L 2 norm on the data gradient as a penalty term. This however acts as a low pass filter and thus is not good at preserving edges in the reconstructed data. In this paper we propose a novel approach whereby an anisotropic regularization is used to preserve object edges. This is achieved by calculating the data gradient over a Riemannian manifold instead of the standard Euclidean space using the Laplace-Beltrami approach. We also employ a modified fidelity term to handle impulse noise. This approach is applicable to both scalar and vector valued images. The result is demonstrate via the Wiener filter with several approaches for minimizing the functional including a novel GSVD based spectral approach applicable to functionals containing gradient based features.