Next: Acknowledgments
Up: Valenciano and Brown: Edge-preserving
Previous: Gradient magnitude and smooth
Introducing an edge-preserving regularization helps to take into account prior knowledge about letter statistics into the least-squares deblurring problem.
The proposed gradient magnitude and Laplacian regularization was the better option for getting rid of the noise and preserving the round sharp features present in the original model.
Stanford Exploration Project
10/14/2003