#include <ILysakerNthOrderOp.hpp>
Inheritance diagram for ILysakerNthOrderOp:
Purpose: Better non-linear preservation of details in TV norme restoration. Paper: "Noise Removal using Fourth-Order Partial Differential Equations with Applications to Medical Magnetic Resonance Images in Space and time.", Marius Lysaker & Arvid Lundervold , Report 2002, UCLA. Purpose: Second order filter like the Rudin/Osher/Fatemi filter are very good denoiser. But is some application, to much details are destroyed. Also, second order filters introduce a staircase effect in images. (like linear steps). Higher order filters can better maintain non-linear information and details, this happens at the cost of maintaining some noise. This is a tradeoff, your choice, or perhaps we may need a kind of "linearity" indicator to decide what type of local restoration should be used.
@ Copyrights: Bernard De Cuyper & Eddy Fraiha 2002, Eggs & Pictures. MIT/Open BSD copyright model.
Public Methods | |
ILysakerNthOrderOp (int iter=50, int im=2, double rdt=0.2, bool colorFlag=false) | |
virtual | ~ILysakerNthOrderOp () |
virtual void | report (FILE *file) |
Protected Methods | |
virtual AnImage * | greyFilter (AnImage *src, AnImage *dest=0) |
virtual AnImage * | greySelectedFilter (AnImage *mask, AnImage *src, AnImage *dest=0) |
virtual Real3CImage * | scolorFilter (Real3CImage *src, Real3CImage *dest=0) |
virtual Real3CImage * | scolorSelectedFilter (AnImage *mask, Real3CImage *src, Real3CImage *dest=0) |
virtual Real3CImage * | fcolorFilter (Real3CImage *src, Real3CImage *dest=0) |
virtual Real3CImage * | fcolorSelectedFilter (AnImage *mask, Real3CImage *src, Real3CImage *dest=0) |
Private Attributes | |
int | m |
double | dt |