Commit 123e14a3 authored by Andrey Filippov's avatar Andrey Filippov

Added initial z0 estimate, switched squared parameters to exp()

parent b2797800
......@@ -96,7 +96,7 @@ public class FocusingField {
private boolean [] rslt_show_chn;
// not saved/restored
private double [] z0_estimates=null; // initial estimates for z0 from the lowest value on data
private double lambdaStepUp; // multiply lambda by this if result is worse
private double lambdaStepDown; // multiply lambda by this if result is better
private double thresholdFinish; // (copied from series) stop iterations if 2 last steps had less improvement (but not worsening )
......@@ -170,6 +170,7 @@ public class FocusingField {
public void setDefaults(){
z0_estimates=null;
sagittalMaster=false; // center data is the same, when true sagittal fitting only may change r=0 coefficients,
parallelOnly = true; // only process measurements for parallel moves
filterInput = true;
......@@ -666,7 +667,12 @@ private boolean [] filterConcave(
int numFilteredInsufficient = 0;
int numFiltered = 0;
int [][] numPoints=new int [getNumChannels()][getNumSamples()];
for (int chn=0;chn<numPoints.length;chn++) for (int sample=0;sample<numPoints[chn].length;sample++) numPoints[chn][sample]=0;
double [][][] z0EstData=new double[getNumChannels()][getNumSamples()][2];
for (int chn=0;chn<numPoints.length;chn++) for (int sample=0;sample<numPoints[chn].length;sample++) {
numPoints[chn][sample]=0;
z0EstData[chn][sample][0]=0.0;
z0EstData[chn][sample][1]=0.0;
}
for (int index=0;index<dataVector.length;index++) if ((index>=enable_in.length) ||enable_in[index]){
numPoints[dataVector[index].channel][dataVector[index].sampleIndex]++;
}
......@@ -811,14 +817,31 @@ private boolean [] filterConcave(
", slope="+ IJ.d2s(100*point_slope[i],3)+ ", concave="+(nonConcave[i]?0.0:1.0));
}
}
// contribute to z0 calculation
for (int i=0;i<thisIndices.length;i++) if (!nonConcave[i]){
z0EstData[chn][sample][1]+=dataWeights[thisIndices[i]];
}
z0EstData[chn][sample][0]=point_z[minIndex];
}
}
if (debugLevel>0) System.out.println("filterConcave(): removed for too few points "+numFilteredInsufficient+" samples");
if (debugLevel>0) System.out.println("filterConcave(): removed for non-concave "+numFiltered+" samples");
// for (int chn=0;chn<numPoints.length;chn++) for (int sample=0;sample<numPoints[chn].length;sample++){
z0_estimates=new double[getNumChannels()];
for (int chn=0;chn<z0_estimates.length;chn++){
double z=0;
double w=0;
for (int sample=0;sample<numPoints[chn].length;sample++){
z+=z0EstData[chn][sample][0]*z0EstData[chn][sample][1];
w+=z0EstData[chn][sample][1];
}
z0_estimates[chn]= (w>0.0)?z/w:Double.NaN;
}
return enable_out;
}
private boolean [] filterTooFar(double ratio,boolean [] enable_in){
if (enable_in==null) {
enable_in=new boolean [dataVector.length];
......@@ -2707,7 +2730,7 @@ public boolean LevenbergMarquardt(boolean openDialog, int debugLevel){
fieldFitting.initSampleCorrVector(
flattenSampleCoord(), //double [][] sampleCoordinates,
getSeriesWeights()); //double [][] sampleSeriesWeights);
fieldFitting.setEstimatedZ0( z0_estimates, false); // boolean force)
this.savedVector=this.fieldFitting.createParameterVector(sagittalMaster);
if (debugDerivativesFxDxDy){
......@@ -3774,6 +3797,18 @@ if ((chn==3) && (sampleIndex==23)){
}
}
public void setEstimatedZ0(
double [] z0,
boolean force){
if ((z0==null)|| (curvatureModel==null)) return; // no estimation available
for (int chn=0;chn<curvatureModel.length;chn++){
if (!Double.isNaN(z0[chn]) && (!curvatureModel[chn].z0IsValid() || force)){
curvatureModel[chn].set_z0(z0[chn]);
if (debugLevel>1) System.out.println("Setting initial (estimated) best focal position for channel "+chn+" = "+z0[chn]);
}
}
}
/**
* create matrix of weights of the other parameters influence
* @param sampleCoordinates [sample number]{x,y} - flattened array of sample coordinates
......@@ -3828,29 +3863,6 @@ if ((chn==3) && (sampleIndex==23)){
}
}
getCorrVector();
/*
sampleCorrChnParIndex=new int [sampleCorrSelect.length][];
int numPars=0;
for (int nChn=0; nChn< sampleCorrCrossWeights.length;nChn++) {
if (sampleCorrCrossWeights[nChn]!=null){
sampleCorrChnParIndex[nChn]=new int [sampleCorrCrossWeights[nChn].length];
for (int nPar=0;nPar< sampleCorrCrossWeights[nChn].length;nPar++) {
if (sampleCorrCrossWeights[nChn][nPar]!=null){
sampleCorrChnParIndex[nChn][nPar]=numPars; // pointer to the first sample
numPars+=sampleCorrCrossWeights[nChn][nPar].length;
} else {
sampleCorrChnParIndex[nChn][nPar]=-1;
}
}
} else {
sampleCorrChnParIndex[nChn]=null;
}
}
// currently all correction parameters are initialized as zeros.
getCorrVector();
if (debugLevel>1) System.out.println("was resetting sampleCorrVector here");
*/
}
......@@ -3888,15 +3900,6 @@ if ((chn==3) && (sampleIndex==23)){
return sampleCoordinates;
}
public double [] getCorrPar(int chn, int sampleIndex){
/* if ((sampleCorrChnParIndex==null) || (sampleCorrChnParIndex[chn]==null)) return null;
double [] corr =new double [sampleCorrChnParIndex[chn].length];
for (int i=0;i<corr.length;i++){
if (sampleCorrChnParIndex[chn][i]<0) corr[i]=0.0;
else corr[i]=sampleCorrVector[sampleCorrChnParIndex[chn][i]+sampleIndex];
}
*/
// System.out.println("used sampleCorrVector here, now correctionParameters");
if (correctionParameters[chn]==null) return null;
double [] corr =new double [correctionParameters[chn].length];
for (int i=0;i<corr.length;i++){
......@@ -4859,8 +4862,8 @@ if ((chn==3) && (sampleIndex==23)){
}
public class CurvatureModel{
private double dflt_na=0.15; // um/um
private double dflt_r0=4.0; //3.3; // um (1.5 pix)
private double dflt_na=Math.log(0.15); // um/um
private double dflt_r0=Math.log(4.0); //3.3; // um (1.5 pix)
private double [][] modelParams=null;
public static final int dflt_distanceParametersNumber=5;
public static final int dflt_radialParametersNumber=4;
......@@ -4934,7 +4937,16 @@ if ((chn==3) && (sampleIndex==23)){
}
this.modelParams[1][0]=dflt_na;
this.modelParams[2][0]=dflt_r0;
this.modelParams[0][0]=Double.NaN;
}
public boolean z0IsValid(){
return !Double.isNaN(this.modelParams[0][0]);
}
public void set_z0(double z0){
this.modelParams[0][0]=z0;
}
public double [] getCenterVector(){
double [] vector=new double [this.modelParams.length];
for (int i=0;i<this.modelParams.length;i++){
......@@ -5006,7 +5018,7 @@ if ((chn==3) && (sampleIndex==23)){
double z_in,
double [] deriv){
/*
f=sqrt((a*(zin-z0))^2 + r0^2)+a0+ a1*(zin-z0)+...aN*(zin-z0)^N
f=sqrt(( a*(zin-z0))^2 + r0^2)+a0+ a1*(zin-z0)+...aN*(zin-z0)^N
each of the z0,z0,a,a[i] is polynomial of even powers of r (r^0, r^2, r^4...)
z=z_in-z0
......@@ -5014,12 +5026,24 @@ modified parameters, r0 - PSF FWHM at z=0, k (instead of a0), so that old r0 now
r0*exp(-k), old a0= r0*(1-exp(-k)).
f=sqrt((a*(zin-z0))^2 + (r0*(exp(-k))^2)+r0*(1-exp(-k))+ a1*(zin-z0)+...aN*(zin-z0)^N
z0 - ar[0]
a - ar[1]
r0 - ar[2]
k - ar[3]
ar1 - ar[4]
modified to avoid zero-crossing for a and r0:
a=exp(ln_a)
r0=exp(ln_r0)
z0 - ar[0]
ln_a - ar[1]
ln_r0 - ar[2]
k - ar[3]
ar1 - ar[4]
*/
double r=Double.isNaN(pY)?pX:Math.sqrt((pX-pX0)*(pX-pX0)+(pY-pY0)*(pY-pY0))*PIXEL_SIZE; // in mm
// double r2=r*r;
......@@ -5035,20 +5059,19 @@ ar1 - ar[4]
ar[i]+=this.modelParams[i][j]*rp;
}
}
double exp_a=Math.exp(ar[1]);
double exp_r=Math.exp(ar[2]);
double z=z_in-ar[0];
double exp=Math.exp(-ar[3]);
double reff=ar[2]*exp;
// double sqrt=Math.sqrt((ar[1]*z)*(ar[1]*z) + ar[2]*ar[2]);
double sqrt=Math.sqrt((ar[1]*z)*(ar[1]*z) + reff*reff);
double exp_k=Math.exp(-ar[3]);
// double reff=ar[2]*exp_k;
double reff=exp_r*exp_k;
// double sqrt=Math.sqrt((ar[1]*z)*(ar[1]*z) + reff*reff);
double sqrt=Math.sqrt((exp_a*z)*(exp_a*z) + reff*reff);
double f=sqrt+ar[2]*(1-exp);
// double f=sqrt+ar[2]*(1-exp_k);
double f=sqrt+exp_r*(1-exp_k);
double zp=1.0;
/*
for (int i=3;i<ar.length;i++){
f+=ar[i]*zp;
zp*=z;
}
*/
for (int i=4;i<ar.length;i++){
zp*=z;
f+=ar[i]*zp;
......@@ -5057,29 +5080,35 @@ ar1 - ar[4]
if (deriv==null) return f; // only value, no derivatives
double [] df_da=new double[this.modelParams.length]; // last element - derivative for dz
// derivative for z0 (shift) - ar[0]
df_da[0]=-1.0/sqrt*ar[1]*ar[1]*z;
// df_da[0]=-1.0/sqrt*ar[1]*ar[1]*z;
df_da[0]=-1.0/sqrt*exp_a*exp_a*z;
zp=1.0;
for (int i=4;i<this.modelParams.length;i++){
df_da[0]-=ar[i]*(i-3)*zp; // ar[i] calculated coefficients for current radius
zp*=z;
}
// derivative for a (related to numeric aperture) - ar[1]
df_da[1]=1.0/sqrt*ar[1]*z*z;
// df_da[1]=1.0/sqrt*ar[1]*z*z;
df_da[1]=1.0/sqrt*exp_a*z*z *exp_a; // d(f)/d(exp_a) *exp_a
// derivative for a (related to lowest PSF radius) - ar[2]
// df_da[2]=1.0/sqrt*ar[2];
df_da[2]=1.0/sqrt*reff*exp + (1-exp); // * exp(-k)
// df_da[2]=1.0/sqrt*reff*exp_k + (1-exp_k); // * exp(-k)
df_da[2]=(1.0/sqrt*reff*exp_k + (1-exp_k)) * exp_r; // d(f)/d(exp_r) *exp_r
// derivative for k (ar[3]
df_da[3]=1.0/sqrt*reff*ar[2]*exp*(-1) + ar[2]*exp;
// df_da[3]=1.0/sqrt*reff*ar[2]*exp_k*(-1) + ar[2]*exp_k;
df_da[3]=1.0/sqrt*reff*exp_r*exp_k*(-1) + exp_r*exp_k;
// derivatives for rest (polynomial) coefficients
zp=1.0;
/*
for (int i=3;i<this.modelParams.length;i++){
df_da[i]=zp;
zp*=z;
}
*/
for (int i=4;i<this.modelParams.length;i++){
zp*=z;
df_da[i]=zp;
......@@ -5089,7 +5118,6 @@ ar1 - ar[4]
double [] dar= new double [this.modelParams[0].length];
dar[0]=1;
for (int j=1;j<dar.length;j++){
// dar[j]=dar[j-1]*r2;
dar[j]=dar[j-1]*r;
if (j==1) dar[j]*=r; // 0,2,3,4,5...
}
......@@ -5120,14 +5148,16 @@ ar1 - ar[4]
}
public String getZName(int i){
if (i==0) return "z0";
if (i==1) return "na";
if (i==2) return "r0";
if (i==1) return "ln(na)";
if (i==2) return "ln(r0)";
if (i==3) return "ln(k)";
else return "az_"+(i-3);
}
public String getZDescription(int i){
if (i==0) return "Focal shift";
if (i==1) return "Defocus/focus shift (~NA)";
if (i==2) return "Best PSF radius";
if (i==1) return "Defocus/focus shift (~NA), ln()";
if (i==2) return "Best PSF radius, ln()";
if (i==3) return "cross shift, ln()";
else return "Polynomial coefficient for z^"+(i-3);
}
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment