@Override
protected OperationData process(IDataset input, IMonitor monitor) {
RectangularROI box = model.getBox();
Dataset in1 =
BoxSlicerRodScanUtils.rOIBox(input, monitor, box.getIntLengths(), box.getIntPoint());
if (g2 == null) g2 = new Polynomial2D(model.getFitPower());
if ((int) Math.pow(model.getFitPower() + 1, 2) != g2.getNoOfParameters())
g2 = new Polynomial2D(model.getFitPower());
Dataset[] fittingBackground =
BoxSlicerRodScanUtils.LeftRightTopBottomBoxes(
input, monitor, box.getIntLengths(), box.getIntPoint(), model.getBoundaryBox());
Dataset offset = DatasetFactory.ones(fittingBackground[2].getShape(), Dataset.FLOAT64);
Dataset intermediateFitTest = Maths.add(offset, fittingBackground[2]);
Dataset matrix =
LinearLeastSquaresServicesForSXRD.polynomial2DLinearLeastSquaresMatrixGenerator(
model.getFitPower(), fittingBackground[0], fittingBackground[1]);
DoubleDataset test = (DoubleDataset) LinearAlgebra.solveSVD(matrix, intermediateFitTest);
double[] params = test.getData();
DoubleDataset in1Background =
g2.getOutputValues0(
params, box.getIntLengths(), model.getBoundaryBox(), model.getFitPower());
IndexIterator it = in1Background.getIterator();
while (it.hasNext()) {
double v = in1Background.getElementDoubleAbs(it.index);
if (v < 0) in1Background.setObjectAbs(it.index, 0);
}
Dataset pBackgroundSubtracted = Maths.subtract(in1, in1Background, null);
pBackgroundSubtracted.setName("pBackgroundSubtracted");
IndexIterator it1 = pBackgroundSubtracted.getIterator();
while (it1.hasNext()) {
double q = pBackgroundSubtracted.getElementDoubleAbs(it1.index);
if (q < 0) pBackgroundSubtracted.setObjectAbs(it1.index, 0);
}
Dataset output = DatasetUtils.cast(pBackgroundSubtracted, Dataset.FLOAT64);
output.setName("Region of Interest, polynomial background removed");
return new OperationData(output);
}
/**
* - * @param axes specify real and imaginary coordinates as two 1d datasets
*
* @return a list containing a dataset of recursion limits
*/
@Override
public List<Dataset> value(IDataset... axes) {
Dataset xaxis, yaxis;
if (axes.length < 2) {
throw new IllegalArgumentException("Need two axes");
}
xaxis = DatasetUtils.convertToDataset(axes[0]);
yaxis = DatasetUtils.convertToDataset(axes[1]);
if (xaxis.getRank() != 1 || yaxis.getRank() != 1) {
throw new IllegalArgumentException("Need both axes to be 1d datasets");
}
IntegerDataset count =
DatasetFactory.zeros(IntegerDataset.class, yaxis.getShapeRef()[0], xaxis.getShapeRef()[0]);
List<Dataset> result = new ArrayList<Dataset>();
final IndexIterator iy = yaxis.getIterator();
int n = 0;
while (iy.hasNext()) {
final double iv = yaxis.getElementDoubleAbs(iy.index);
final IndexIterator ix = xaxis.getIterator();
while (ix.hasNext()) {
final double rv = xaxis.getElementDoubleAbs(ix.index);
double x = 0, y = 0;
int c = -1;
do {
double t = x * x - y * y + rv;
y = 2. * x * y + iv;
x = t;
} while (++c < maxRecursion && x * x + y * y <= 4.);
count.setAbs(n++, c);
}
}
result.add(count);
return result;
}
/**
* Fast statistics as a rough guide - this is faster than Dataset.getMin() and getMax() which may
* cache but slows the opening of images too much. The return array[2] was added in "Updated for
* Diffraction Tool." commit, but no trace of such usage. However it should not be removed,
* because it is useful as return array[3].
*
* @param bean
* @return [0] = min [1] = max(=mean*constant) [2] = mean [3] max
*/
public double[] getFastStatistics(ImageServiceBean bean) {
Dataset image = getImageLoggedData(bean);
if (bean.getHistogramType() == HistoType.OUTLIER_VALUES && !bean.isLogColorScale()) {
double[] ret = null;
try {
double[] stats = Stats.outlierValues(image, bean.getLo(), bean.getHi(), -1);
ret = new double[] {stats[0], stats[1], -1};
} catch (IllegalArgumentException iae) {
bean.setLo(10);
bean.setHi(90);
double[] stats = Stats.outlierValues(image, bean.getLo(), bean.getHi(), -1);
ret = new double[] {stats[0], stats[1], -1};
}
if (bean.isLogColorScale() && ret != null) {
ret = new double[] {Math.pow(10, ret[0]), Math.pow(10, ret[1]), -1};
}
return ret;
}
double min = Double.MAX_VALUE;
double max = -Double.MAX_VALUE;
double sum = 0.0;
int size = 0;
BooleanDataset mask =
bean.getMask() != null
? (BooleanDataset) DatasetUtils.cast(bean.getMask(), Dataset.BOOL)
: null;
// Big loop warning:
final IndexIterator it = image.getIterator();
final IndexIterator mit = mask == null ? null : mask.getIterator();
while (it.hasNext()) {
final double val = image.getElementDoubleAbs(it.index);
if (mit != null && mit.hasNext()) {
if (!mask.getElementBooleanAbs(mit.index)) {
continue; // Masked!
}
}
if (Double.isNaN(val)) continue;
if (!bean.isInBounds(val)) continue;
sum += val;
if (val < min) min = val;
if (val > max) max = val;
size++;
}
double retMax = Double.NaN;
double retExtra = Double.NaN;
if (bean.getHistogramType() == HistoType.MEDIAN) {
double median = Double.NaN;
try {
median = ((Number) Stats.median(image)).doubleValue(); // SLOW
} catch (Exception ne) {
median = ((Number) Stats.median(image.cast(Dataset.INT16))).doubleValue(); // SLOWER
}
retMax = 2 * median;
retExtra = median;
} else { // Use mean based histo
double mean = sum / size;
retMax = (Math.E) * mean; // Not statistical, E seems to be better than 3...
retExtra = mean;
}
if (retMax > max) retMax = max;
if (bean.isLogColorScale()) {
return new double[] {Math.pow(10, min), Math.pow(10, retMax), Math.pow(10, retExtra)};
}
return new double[] {min, retMax, retExtra, max};
}
