---

FClassicAnisotropicSpace Class Reference

#include <FClassicAnisotropicSpace.hpp>

Inheritance diagram for FClassicAnisotropicSpace:

[legend]Collaboration diagram for FClassicAnisotropicSpace:

[legend]List of all members.

---

Detailed Description

Classic Anisotropic Spatial Discretisation.

Author:

Bernard De Cuyper

Version:

1.15

Date:

14/08/2004

Remarks:

Tornado1 Spatial discretisation

Nonlinear Diffusion

Convex Potential

Remarks:

3.315 is essential, the flux increase when s is smaller than the contrast (denoising) and the flux decreases when s is greater than the contrast (enhancing)

Other Convex Potentials have been proposed:
1. Deriche/Santana: g(s)= exp(-s^2/contrast^2)
2. Perona-Malik: g(s)= 1.0/ (1.0 + s^2/contrast^2)
3. Total Variation: g(s)= 1.0/ sqrt( eps^2 + s^2/contrast^2 ) 0< eps <<1.0
4. Weickert: g(s2)= 1.0 - exp(Cm/(s2/contrast2)^m ) if s2>0.0 , m={2,3,4}

 
Purpose:        2D Classic Anisotropic Discretisation used in nxn Linear Solver
                        Used to generate a 2D Classic Anisotropic Discretisation,....

                Nonlinear Diffusion tend to segment or flat images,(However Lysaker proves non linearity can be obtained)
                It has the advantage to soft regions, and to contrast/enhance edges.
                It is less recommanded in strong noise images.

Paper:  "Efficient Image Segmentation Using Partial Differential Equations and Morphology.",
                Joachim Weickert, Technical Report 3/2000, Computer Science Series, 
                DIKU, University of Copenhagen, Denmark.

                "Evaluation of Diffusion Schemes for Multi-scale Watershed Segmentation.", Erik B. Dam,
                Thesis 2000, Computer Science, University of Copenhagen.

                "Edge-Preseving Noise Removal, Part I: Second-Order Anisotropic Diffusion.",
                Seongjai Kim, Technical Report 2001-09, University of Kentucky.

                "MinBAD: The Minimum-Biased Anisotropic Diffusion for Noise Removal.",
                Seongjai Kim & all, Technical Report 2002-06, University of Kentucky.

                "PDE-based image restoration,II: Numerical schemes and color image denoising.",
                Seongjai Kim & all, Technical Report 2003-08, University of Kentucky.

                "Image Analysis & Edge-Preseving Noise Removal.",
                Seongjai Kim, Presentation 2003, University of Kentucky.
  

                

@ Copyrights: Bernard De Cuyper 2004, Eggs & Pictures. MIT/Open BSD copyright model.

Public Methods
	FClassicAnisotropicSpace (double sigma=2.0, double acontrast=1.0, int diffusivity=0, int gradType=0)
virtual	~FClassicAnisotropicSpace ()
virtual void	init (FImage *uk)
virtual AFSymMatrix *	getA (float tau, FImage *uk, AFSymMatrix *A=0)
virtual AFSymMatrix *	getRowA (float tau, FImage *uk, int row, AFSymMatrix *A1=0)
virtual AFSymMatrix *	getColA (float tau, FImage *uk, int col, AFSymMatrix *A2=0)
Protected Methods
virtual FImage *	gValues (FImage *src, FImage *dest=0)
Protected Attributes
float	sigmaRegularisation
float	contrast
float	contrast2
int	diffusivity
AbsFImageFilter *	blurrer
FGradientNormOp *	gradient
FImage *	uSigma
FImage *	g

---

The documentation for this class was generated from the following files:

• FClassicAnisotropicSpace.hpp
• FClassicAnisotropicSpace.cpp

Restoreinpaint sourceforge project `C++/Java Image Processing, Restoration, Inpainting Project'.

Bernard De Cuyper: Open Project Leader: Concept, design and development. Bernard De Cuyper & Eddy Fraiha 2002, 2003. Bernard De Cuyper 2004. Open and free, for friendly usage only. Modifications on Belgium ground of this piece of artistic work, by governement institutions or companies, must be notified to Bernard De Cuyper.