Utils (Fiji Javadocs 0-SNAPSHOT API)

java.lang.Object
- trainableSegmentation.utils.Utils

```
public final class Utils
extends Object
```
This class implements useful methods for the Weka Segmentation library.

Method Summary

All Methods Static Methods Concrete Methods
Modifier and Type	Method and Description
`static ij.measure.ResultsTable`	`confusionMatrix(ij.ImagePlus prediction, ij.ImagePlus groundtruth, ArrayList<String> classes, int[] classIndexToLabel)` Calculate confusion matrix bewtween two label images.
`static ij.measure.ResultsTable`	`confusionMatrix(ij.process.ImageProcessor prediction, ij.process.ImageProcessor groundtruth, ArrayList<String> classes, int[] classIndexToLabel)` Calculate confusion matrix bewtween two label images.
`static ij.measure.ResultsTable`	`confusionMatrix(ij.ImageStack prediction, ij.ImageStack groundtruth, ArrayList<String> classes, int[] classIndexToLabel)` Calculate confusion matrix bewtween two label images.
`static FindConnectedRegions.Results`	`connectedComponents(ij.ImagePlus im, int adjacency)` Connected components based on Find Connected Regions (from Mark Longair)
`static FindConnectedRegions.Results`	`connectedComponents(ij.ImagePlus im, int adjacency, int minSize)` Connected components based on Find Connected Regions (from Mark Longair)
`static Plot`	`createPrecisionRecallPlot(ArrayList<ClassificationStatistics> stats)` Create plot with the precision-recall curve
`static void`	`dilate(ij.process.FloatProcessor fp)` Applies the morphological dilate operator.
`static void`	`erode(ij.process.FloatProcessor fp)` Applies the morphological erode operator.
`static void`	`fill(ij.process.ImageProcessor ip, int foreground, int background)` Binary fill
`static void`	`filterSmallObjectsAndHoles(ij.process.FloatProcessor probabilityMap, double thresholdValue, int minSize)` Filter small objects and holes at a specific threshold value
`static ArrayList<Point3f>[]`	`getClassCoordinates(ij.ImagePlus labelImage, ij.ImagePlus mask)` Get the binary class coordinates from a label image (2D image or stack)
`static ij.process.LUT`	`getGoldenAngleLUT()` Create golden angle LUT (starting in red)
`static double`	`getKappa(ClassificationStatistics stats)` Get Kappa statistic
`static double`	`getPrecRecArea(ArrayList<ClassificationStatistics> stats)` Get area under the Precision/Recall curve
`static double`	`getROCArea(ArrayList<ClassificationStatistics> stats)` Calculates the area under the ROC curve as the Wilcoxon-Mann-Whitney statistic.
`static boolean`	`insertImage(ij.ImagePlus src, ij.ImagePlus dst, int[] origin)` Insert an image into another one (2D or 3D).
`static ij.ImagePlus[]`	`maxPool(ij.ImagePlus input, ij.ImagePlus label, int sizeX, int sizeY)` Experimental max pooling method.
`static ij.ImagePlus[]`	`maxPoolNoReduction(ij.ImagePlus input, ij.ImagePlus label, int sizeX, int sizeY)` Experimental max pooling method without size reduction.
`static ij.ImageStack`	`normalize(ij.ImageStack inputStack)` Normalize an image stack so it has 0 mean and unit variance
`static void`	`normalize01(ij.process.FloatProcessor fp)` Normalize float image so the pixel are between 0 and 1
`static void`	`plotPrecisionRecall(ArrayList<ClassificationStatistics> stats)` Plot the precision-recall curve
`static void`	`plotROC(ArrayList<ClassificationStatistics> stats)` Plot the Receiver operating characteristic curve
`static void`	`postProcess(ij.process.FloatProcessor probabilityMap, int smoothIterations, double threshold, int minSize, boolean binarize)` Post-process probability image to get more reasonable objects at a certain threshold
`static void`	`smooth(ij.process.FloatProcessor fp, float f1, float f2, float f3)` Blurs probability image with a given symmetrically weighted kernel.
`static ij.process.ByteProcessor`	`threshold(ij.process.ImageProcessor ip, double thresholdValue)` Apply binary threshold to input image

Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

- Method Detail
  - connectedComponents
```
public static FindConnectedRegions.Results connectedComponents(ij.ImagePlus im,
                                                               int adjacency)
```
    Connected components based on Find Connected Regions (from Mark Longair)
    
    Parameters:
    
    im - input image
    
    adjacency - number of neighbors to check (4, 8...)
    
    Returns:
    
    list of images per region, all-regions image and regions info
  - connectedComponents
```
public static FindConnectedRegions.Results connectedComponents(ij.ImagePlus im,
                                                               int adjacency,
                                                               int minSize)
```
    Connected components based on Find Connected Regions (from Mark Longair)
    
    Parameters:
    
    im - input image
    
    adjacency - number of neighbors to check (4, 8...)
    
    minSize - minimum size (in pixels) of the components
    
    Returns:
    
    list of images per region, all-regions image and regions info
  - plotPrecisionRecall
```
public static void plotPrecisionRecall(ArrayList<ClassificationStatistics> stats)
```
    Plot the precision-recall curve
    
    Parameters:
    
    stats - classification statistics
  - plotROC
```
public static void plotROC(ArrayList<ClassificationStatistics> stats)
```
    Plot the Receiver operating characteristic curve
    
    Parameters:
    
    stats - classification statistics
  - getPrecRecArea
```
public static double getPrecRecArea(ArrayList<ClassificationStatistics> stats)
```
    Get area under the Precision/Recall curve
    
    Parameters:
    
    stats - classification statistics with the ROC curve information
    
    Returns:
    
    area under the input curve
  - getROCArea
```
public static double getROCArea(ArrayList<ClassificationStatistics> stats)
```
    Calculates the area under the ROC curve as the Wilcoxon-Mann-Whitney statistic.
    
    Parameters:
    
    stats - classification statistics .
    
    Returns:
    
    the ROC area, or Double.NaN if you don't pass in a ThresholdCurve generated Instances.
  - getKappa
```
public static double getKappa(ClassificationStatistics stats)
```
    Get Kappa statistic
    
    Parameters:
    
    stats - classification statistics
    
    Returns:
    
    Kappa statistic
  - createPrecisionRecallPlot
```
public static Plot createPrecisionRecallPlot(ArrayList<ClassificationStatistics> stats)
```
    Create plot with the precision-recall curve
    
    Parameters:
    
    stats - classification statistics
    
    Returns:
    
    precision-recall plot
  - normalize
```
public static ij.ImageStack normalize(ij.ImageStack inputStack)
```
    Normalize an image stack so it has 0 mean and unit variance
    
    Parameters:
    
    inputStack - input stack
    
    Returns:
    
    normalize stack
  - smooth
```
public static void smooth(ij.process.FloatProcessor fp,
                          float f1,
                          float f2,
                          float f3)
```
    Blurs probability image with a given symmetrically weighted kernel. Reused from the SIOX package.
    
    Parameters:
    
    fp - probability image to be smoothed
    
    f1 - Weight factor for the first pixel.
    
    f2 - Weight factor for the mid-pixel.
    
    f3 - Weight factor for the last pixel.
  - erode
```
public static void erode(ij.process.FloatProcessor fp)
```
    Applies the morphological erode operator. Reused from the SIOX package.
    
    Parameters:
    
    fp - probability image to be eroded.
  - dilate
```
public static void dilate(ij.process.FloatProcessor fp)
```
    Applies the morphological dilate operator. Reused from the SIOX package. Can be used to close small holes in the probability image.
    
    Parameters:
    
    fp - probability image to be dilated
  - threshold
```
public static ij.process.ByteProcessor threshold(ij.process.ImageProcessor ip,
                                                 double thresholdValue)
```
    Apply binary threshold to input image
    
    Parameters:
    
    ip - input image
    
    thresholdValue - threshold value (all pixel above that value will be set to 255, the rest to 0)
    
    Returns:
    
    binary result
  - postProcess
```
public static void postProcess(ij.process.FloatProcessor probabilityMap,
                               int smoothIterations,
                               double threshold,
                               int minSize,
                               boolean binarize)
```
    Post-process probability image to get more reasonable objects at a certain threshold
    
    Parameters:
    
    probabilityMap - probability image
    
    smoothIterations - number of smoothing iterations
    
    threshold - threshold to use
    
    minSize - minimum object size (in pixels)
    
    binarize - flag to binarize results
  - normalize01
```
public static void normalize01(ij.process.FloatProcessor fp)
```
    Normalize float image so the pixel are between 0 and 1
    
    Parameters:
    
    fp - input image
  - filterSmallObjectsAndHoles
```
public static void filterSmallObjectsAndHoles(ij.process.FloatProcessor probabilityMap,
                                              double thresholdValue,
                                              int minSize)
```
    Filter small objects and holes at a specific threshold value
    
    Parameters:
    
    probabilityMap - probability image
    
    thresholdValue - threshold to use
    
    minSize - minimum size of the objects (in pixels)
  - fill
```
public static void fill(ij.process.ImageProcessor ip,
                        int foreground,
                        int background)
```
    Binary fill
    
    Parameters:
    
    ip - input image
    
    foreground - foreground value
    
    background - background value
  - getClassCoordinates
```
public static ArrayList<Point3f>[] getClassCoordinates(ij.ImagePlus labelImage,
                                                       ij.ImagePlus mask)
```
    Get the binary class coordinates from a label image (2D image or stack)
    
    Parameters:
    
    labelImage - labels (they can be in any format, black = 0)
    
    mask - binary mask to select the pixels to be extracted
    
    Returns:
    
    array with the two lists (black and white) of sample coordinates
  - maxPool
```
public static ij.ImagePlus[] maxPool(ij.ImagePlus input,
                                     ij.ImagePlus label,
                                     int sizeX,
                                     int sizeY)
```
    Experimental max pooling method.
    
    Parameters:
    
    input - input image
    
    label - label image
    
    sizeX - width of max pooling filter
    
    sizeY - height of max pooling filter
    
    Returns:
    
    input and label images after max pooling
  - maxPoolNoReduction
```
public static ij.ImagePlus[] maxPoolNoReduction(ij.ImagePlus input,
                                                ij.ImagePlus label,
                                                int sizeX,
                                                int sizeY)
```
    Experimental max pooling method without size reduction.
    
    Parameters:
    
    input - input image
    
    label - label image
    
    sizeX - width of max pooling filter
    
    sizeY - height of max pooling filter
    
    Returns:
    
    input and label images after max pooling
  - getGoldenAngleLUT
```
public static ij.process.LUT getGoldenAngleLUT()
```
    Create golden angle LUT (starting in red)
    
    Returns:
    
    golden angle LUT
  - insertImage
```
public static boolean insertImage(ij.ImagePlus src,
                                  ij.ImagePlus dst,
                                  int[] origin)
```
    Insert an image into another one (2D or 3D).
    
    Parameters:
    
    src - source image
    
    dst - destination image
    
    origin - coordinates of the insertion origin in the destination image (zero-based numbering)
    
    Returns:
    
    false if an error occurs, true otherwise
  - confusionMatrix
```
public static ij.measure.ResultsTable confusionMatrix(ij.process.ImageProcessor prediction,
                                                      ij.process.ImageProcessor groundtruth,
                                                      ArrayList<String> classes,
                                                      int[] classIndexToLabel)
```
    Calculate confusion matrix bewtween two label images. The values are returned following Matlab's 'plotconfusion' function. The rows correspond to the predicted class and the columns to the groundtruth class. The diagonal cells correspond to the samples correctly classified while the off-diagonal cells are incorrectly classified samples. Precision (positive predictive value) is displayed in the far right column and recall (true positive rate) is displayed in the bottom row. The cell in the bottom right of the table shows the overall accuracy.
    
    Parameters:
    
    prediction - predicted labels
    
    groundtruth - groundtruth labels
    
    classes - list of class names
    
    classIndexToLabel - correspondence between class index and label values
    
    Returns:
    
    confusion matrix ready to be plotted
  - confusionMatrix
```
public static ij.measure.ResultsTable confusionMatrix(ij.ImageStack prediction,
                                                      ij.ImageStack groundtruth,
                                                      ArrayList<String> classes,
                                                      int[] classIndexToLabel)
```
    Calculate confusion matrix bewtween two label images. The values are returned following Matlab's 'plotconfusion' function. The rows correspond to the predicted class and the columns to the groundtruth class. The diagonal cells correspond to the samples correctly classified while the off-diagonal cells are incorrectly classified samples. Precision (positive predictive value) is displayed in the far right column and recall (true positive rate) is displayed in the bottom row. The cell in the bottom right of the table shows the overall accuracy.
    
    Parameters:
    
    prediction - predicted labels
    
    groundtruth - groundtruth labels
    
    classes - list of class names
    
    classIndexToLabel - correspondence between class index and label values
    
    Returns:
    
    confusion matrix ready to be plotted
  - confusionMatrix
```
public static ij.measure.ResultsTable confusionMatrix(ij.ImagePlus prediction,
                                                      ij.ImagePlus groundtruth,
                                                      ArrayList<String> classes,
                                                      int[] classIndexToLabel)
```
    Calculate confusion matrix bewtween two label images. The values are returned following Matlab's 'plotconfusion' function. The rows correspond to the predicted class and the columns to the groundtruth class. The diagonal cells correspond to the samples correctly classified while the off-diagonal cells are incorrectly classified samples. Precision (positive predictive value) is displayed in the far right column and recall (true positive rate) is displayed in the bottom row. The cell in the bottom right of the table shows the overall accuracy.
    
    Parameters:
    
    prediction - predicted labels
    
    groundtruth - groundtruth labels
    
    classes - list of class names
    
    classIndexToLabel - correspondence between class index and label values
    
    Returns:
    
    confusion matrix ready to be plotted

Class Utils

Method Summary

Methods inherited from class java.lang.Object

Method Detail

connectedComponents

connectedComponents

plotPrecisionRecall

plotROC

getPrecRecArea

getROCArea

getKappa

createPrecisionRecallPlot

normalize

smooth

erode

dilate

threshold

postProcess

normalize01

filterSmallObjectsAndHoles

fill

getClassCoordinates

maxPool

maxPoolNoReduction

getGoldenAngleLUT

insertImage

confusionMatrix

confusionMatrix

confusionMatrix