Page MenuHomec4science
Files F82451325

TikhonovRegularizedInverseFilter.java
No OneTemporary
Actions

File Metadata

Created
Wed, Sep 11, 12:11

TikhonovRegularizedInverseFilter.java
View Options

/*
* DeconvolutionLab2
*
* Conditions of use: You are free to use this software for research or
* educational purposes. In addition, we expect you to include adequate
* citations and acknowledgments whenever you present or publish results that
* are based on it.
*
* Reference: DeconvolutionLab2: An Open-Source Software for Deconvolution
* Microscopy D. Sage, L. Donati, F. Soulez, D. Fortun, G. Schmit, A. Seitz,
* R. Guiet, C. Vonesch, M Unser, Methods of Elsevier, 2017.
*/
/*
* Copyright 2010-2017 Biomedical Imaging Group at the EPFL.
*
* This file is part of DeconvolutionLab2 (DL2).
*
* DL2 is free software: you can redistribute it and/or modify it under the
* terms of the GNU General Public License as published by the Free Software
* Foundation, either version 3 of the License, or (at your option) any later
* version.
*
* DL2 is distributed in the hope that it will be useful, but WITHOUT ANY
* WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
* A PARTICULAR PURPOSE. See the GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along with
* DL2. If not, see <http://www.gnu.org/licenses/>.
*/
package deconvolution.algorithm;
import java.util.concurrent.Callable;
import signal.ComplexSignal;
import signal.Operations;
import signal.RealSignal;
import signal.SignalCollector;
import signal.factory.complex.ComplexSignalFactory;
public class TikhonovRegularizedInverseFilter extends AbstractAlgorithm implements Callable<RealSignal> {
private double lambda = 0.1;
public TikhonovRegularizedInverseFilter(double lambda) {
super();
this.lambda = lambda;
}
@Override
public RealSignal call() {
if (optimizedMemoryFootprint)
return runOptimizedMemoryFootprint();
else
return runTextBook();
}
public RealSignal runTextBook() {
ComplexSignal Y = fft.transform(y);
ComplexSignal H = fft.transform(h);
ComplexSignal H2 = Operations.multiply(H, H);
ComplexSignal I = ComplexSignalFactory.identity(Y.nx, Y.ny, Y.nz);
I.times((float)lambda);
ComplexSignal FA = Operations.add(H2, I);
ComplexSignal FT = Operations.divideStabilized(H, FA);
ComplexSignal X = Operations.multiply(Y, FT);
RealSignal x = fft.inverse(X);
SignalCollector.free(FT);
SignalCollector.free(Y);
SignalCollector.free(H);
SignalCollector.free(FA);
SignalCollector.free(I);
SignalCollector.free(H2);
SignalCollector.free(X);
return x;
}
public RealSignal runOptimizedMemoryFootprint() {
ComplexSignal Y = fft.transform(y);
ComplexSignal H = fft.transform(h);
ComplexSignal X = filter(Y, H);
SignalCollector.free(H);
SignalCollector.free(Y);
RealSignal x = fft.inverse(X);
SignalCollector.free(X);
return x;
}
private ComplexSignal filter(ComplexSignal Y, ComplexSignal H) {
int nx = H.nx;
int ny = H.ny;
int nz = H.nz;
int nxy = nx * ny*2;
float ya, yb, ha, hb, fa, fb, mag, ta, tb;
float epsilon2 = (float)(Operations.epsilon * Operations.epsilon);
ComplexSignal result = new ComplexSignal("TRIF", nx, ny, nz);
float l = (float)lambda;
for(int k=0; k<nz; k++)
for(int i=0; i< nxy; i+=2) {
ha = H.data[k][i];
hb = H.data[k][i+1];
ya = Y.data[k][i];
yb = Y.data[k][i+1];
fa = ha*ha - hb*hb + l;
fb = 2f * ha * hb;
mag = fa*fa + fb*fb;
ta = (ha*fa + hb*fb) / (mag >= epsilon2 ? mag : epsilon2);
tb = (hb*fa - ha*fb) / (mag >= epsilon2 ? mag : epsilon2);
result.data[k][i] = ya*ta - yb*tb;
result.data[k][i+1] = ya*tb + ta*yb;
}
return result;
}
@Override
public int getComplexityNumberofFFT() {
return 3;
}
@Override
public String getName() {
return "Tikhonov Regularization Inverse Filter";
}
@Override
public String[] getShortnames() {
return new String[] {"TRIF", "TR"};
}
@Override
public double getMemoryFootprintRatio() {
return 8.0;
}
@Override
public boolean isRegularized() {
return true;
}
@Override
public boolean isStepControllable() {
return false;
}
@Override
public boolean isIterative() {
return false;
}
@Override
public boolean isWaveletsBased() {
return false;
}
@Override
public void setParameters(double[] params) {
if (params == null)
return;
if (params.length > 0)
lambda = (float)params[0];
}
@Override
public double[] getDefaultParameters() {
return new double[] {0.1};
}
@Override
public double[] getParameters() {
return new double[] {lambda};
}
@Override
public double getRegularizationFactor() {
return lambda;
}
@Override
public double getStepFactor() {
return 0.0;
}
}

Event Timeline

c4science · Help