Pixel classification refactoring

Major refactoring of pixel classification, including:
- Pixel layer added as an ImageData property, displayed by default with the pixel layer overlay & supporting dynamic measurements on request
- Added 'getOutputType()' for ImageServers in general, to support more server implementations that return classifications/features
- Added (rough) implementation of simple threshold command

qupath-core/src/main/java/qupath/lib/classifiers/pixel/PixelClassificationImageServer.java

@@ -157,7 +157,7 @@ protected BufferedImage readTile(TileRequest tileRequest) throws IOException {
 	public int getClassification(int x, int y, int z, int t) throws IOException {
 
 		var type = classifier.getMetadata().getOutputType();
-		if (type != ImageServerMetadata.OutputType.CLASSIFICATION && type != ImageServerMetadata.OutputType.PROBABILITIES)
+		if (type != ImageServerMetadata.OutputType.CLASSIFICATION && type != ImageServerMetadata.OutputType.PROBABILITY)
 			return -1;
 
 		var tile = getTile(0, x, y, z, t);
@@ -186,7 +186,7 @@ public int getClassification(int x, int y, int z, int t) throws IOException {
 				logger.error("Error requesting classification", e);
 				return -1;
 			}
-		} else if (type == ImageServerMetadata.OutputType.PROBABILITIES) {
+		} else if (type == ImageServerMetadata.OutputType.PROBABILITY) {
 			int maxInd = -1;
 			double maxVal = Double.NEGATIVE_INFINITY;
 			var raster = img.getRaster();

qupath-core/src/main/java/qupath/lib/classifiers/pixel/PixelClassifierMetadata.java

@@ -196,7 +196,7 @@ public static class Builder {
     	private PixelType inputDataType = PixelType.UInt8;
     	private int outputWidth = -1;
     	private int outputHeight = -1;
-    	private ImageServerMetadata.OutputType outputType = ImageServerMetadata.OutputType.PROBABILITIES;
+    	private ImageServerMetadata.OutputType outputType = ImageServerMetadata.OutputType.PROBABILITY;
     	private List<ImageChannel> channels = new ArrayList<>();
 
     	public PixelClassifierMetadata build() {

qupath-core/src/main/java/qupath/lib/images/servers/ImageServer.java

@@ -214,7 +214,7 @@ public interface ImageServer<T> extends AutoCloseable {
 	 * @return
 	 */
 	public default ImageServerMetadata.OutputType getOutputType() {
-		return ImageServerMetadata.OutputType.CHANNELS;
+		return ImageServerMetadata.OutputType.CHANNEL;
 	}
 
 	/**

qupath-core/src/main/java/qupath/lib/images/servers/ImageServerMetadata.java

@@ -45,17 +45,19 @@
  */
 public class ImageServerMetadata {
 
-	public static enum OutputType { CHANNELS, FEATURES, PROBABILITIES, CLASSIFICATION;
+	public static enum OutputType { CHANNEL, FEATURE, PROBABILITY, MULTICLASS_PROBABILITY, CLASSIFICATION;
 
 		@Override
 		public String toString() {
 			switch (this) {
-			case CHANNELS:
-				return "Channels";
-			case FEATURES:
-				return "Features";
-			case PROBABILITIES:
-				return "Probabilities";
+			case CHANNEL:
+				return "Channel";
+			case FEATURE:
+				return "Feature";
+			case PROBABILITY:
+				return "Probability";
+			case MULTICLASS_PROBABILITY:
+				return "Multiclass probability";
 			case CLASSIFICATION:
 				return "Classification";
 			default:
@@ -78,7 +80,7 @@ public String toString() {
 	private int sizeZ = 1;
 	private int sizeT = 1;
 
-	public ImageServerMetadata.OutputType outputType = ImageServerMetadata.OutputType.CHANNELS;
+	public ImageServerMetadata.OutputType outputType = ImageServerMetadata.OutputType.CHANNEL;
 
 	private boolean isRGB = false;
 	private int bitDepth = 8;

qupath-experimental/src/main/java/qupath/lib/classifiers/opencv/gui/PixelClassifierImageSelectionPane.java

@@ -259,7 +259,7 @@ public void initialize(final QuPathViewer viewer) {
 		// Output
 		var labelOutput = new Label("Output");
 		var comboOutput = new ComboBox<ImageServerMetadata.OutputType>();
-		comboOutput.getItems().addAll(ImageServerMetadata.OutputType.CLASSIFICATION, ImageServerMetadata.OutputType.PROBABILITIES);
+		comboOutput.getItems().addAll(ImageServerMetadata.OutputType.CLASSIFICATION, ImageServerMetadata.OutputType.PROBABILITY);
 		selectedOutputType = comboOutput.getSelectionModel().selectedItemProperty();
 		selectedOutputType.addListener((v, o, n) -> {
 			updateClassifier();

qupath-experimental/src/main/java/qupath/lib/classifiers/pixel/OpenCVPixelClassifier.java

@@ -149,7 +149,7 @@ public BufferedImage applyClassification(final ImageData<BufferedImage> imageDat
 
     	ColorModel colorModelLocal = null;
 
-    	if (type == ImageServerMetadata.OutputType.PROBABILITIES) {
+    	if (type == ImageServerMetadata.OutputType.PROBABILITY) {
     		var matProbabilities = matOutput;
     		double maxValue = 1.0;
     		if (do8Bit()) {

qupath-gui-fx/src/main/java/qupath/lib/gui/models/ObservableMeasurementTableData.java

@@ -264,7 +264,7 @@ public synchronized void updateMeasurementList() {
 			var pixelClassifier = imageData.getProperty("PIXEL_LAYER");
 			if (pixelClassifier instanceof ImageServer<?>) {
 				ImageServer<BufferedImage> server = (ImageServer<BufferedImage>)pixelClassifier;
-				if (server.getOutputType() == ImageServerMetadata.OutputType.CLASSIFICATION || server.getOutputType() == ImageServerMetadata.OutputType.PROBABILITIES) {
+				if (server.getOutputType() == ImageServerMetadata.OutputType.CLASSIFICATION || server.getOutputType() == ImageServerMetadata.OutputType.PROBABILITY) {
 					var pixelManager = new PixelClassificationMeasurementManager(server);
 					for (String name : pixelManager.getMeasurementNames()) {
 						builderMap.put(name, new PixelClassifierMeasurementBuilder(pixelManager, name));

qupath-gui-fx/src/main/java/qupath/lib/gui/models/PixelClassificationMeasurementManager.java

@@ -124,7 +124,7 @@ synchronized MeasurementList calculateMeasurements(final ROI roi, final boolean
 
         // Check we have a suitable output type
         ImageServerMetadata.OutputType type = classifierServer.getOutputType();
-        if (type == ImageServerMetadata.OutputType.FEATURES)
+        if (type == ImageServerMetadata.OutputType.FEATURE)
   			return null;
 
 
@@ -204,7 +204,7 @@ synchronized MeasurementList calculateMeasurements(final ROI roi, final boolean
 					logger.error("Error calculating classification areas", e);
 				}
 				break;
-			case PROBABILITIES:
+			case PROBABILITY:
 				// Take classification from the channel with the highest value
 				raster = tile.getRaster();
 				rasterMask = imgMask.getRaster();