/*
* GeoTools - The Open Source Java GIS Toolkit
* http://geotools.org
*
* (C) 2005-2008, Open Source Geospatial Foundation (OSGeo)
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation;
* version 2.1 of the License.
*
* This library 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
* Lesser General Public License for more details.
*/
package org.geotools.renderer.lite.gridcoverage2d;
import java.awt.RenderingHints;
import java.awt.Transparency;
import java.awt.color.ColorSpace;
import java.awt.image.ColorModel;
import java.awt.image.ComponentColorModel;
import java.awt.image.DataBuffer;
import java.awt.image.RenderedImage;
import java.awt.image.renderable.ParameterBlock;
import java.util.Collections;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Set;
import javax.media.jai.Histogram;
import javax.media.jai.ImageLayout;
import javax.media.jai.JAI;
import javax.media.jai.LookupTableJAI;
import javax.media.jai.ParameterBlockJAI;
import javax.media.jai.RenderedOp;
import javax.media.jai.operator.ExtremaDescriptor;
import javax.media.jai.operator.HistogramDescriptor;
import javax.media.jai.operator.LookupDescriptor;
import org.geotools.coverage.GridSampleDimension;
import org.geotools.coverage.grid.GridCoverage2D;
import org.geotools.coverage.grid.GridCoverageFactory;
import org.geotools.coverage.grid.GridGeometry2D;
import org.geotools.factory.Hints;
import org.geotools.image.ImageWorker;
import org.geotools.referencing.piecewise.DefaultPiecewiseTransform1D;
import org.geotools.referencing.piecewise.DefaultPiecewiseTransform1DElement;
import org.geotools.referencing.piecewise.GenericPiecewise;
import org.geotools.referencing.piecewise.MathTransform1DAdapter;
import org.geotools.referencing.piecewise.PiecewiseTransform1D;
import org.geotools.renderer.i18n.ErrorKeys;
import org.geotools.renderer.i18n.Errors;
import org.geotools.renderer.i18n.Vocabulary;
import org.geotools.renderer.i18n.VocabularyKeys;
import org.geotools.resources.image.ColorUtilities;
import org.geotools.styling.ContrastEnhancement;
import org.geotools.styling.StyleVisitor;
import org.geotools.util.NumberRange;
import org.geotools.util.SimpleInternationalString;
import org.opengis.coverage.grid.GridCoverage;
import org.opengis.filter.expression.Expression;
import org.opengis.referencing.operation.TransformException;
import org.opengis.util.InternationalString;
/**
* This implementations of {@link CoverageProcessingNode} takes care of the
* {@link ContrastEnhancement} element of the SLD 1.0 spec.
*
* @author Simone Giannecchini, GeoSolutions
*
*/
class ContrastEnhancementNode extends StyleVisitorCoverageProcessingNodeAdapter
implements StyleVisitor, CoverageProcessingNode {
/**
* Minimal normalized value.
*/
private static final double MIN_VALUE = 0d;
/**
* Maximal normalized value.
*/
private static final double MAX_VALUE = 1d;
/*
* (non-Javadoc)
* @see CoverageProcessingNode#getName()
*/
public InternationalString getName() {
return Vocabulary.formatInternational(VocabularyKeys.CONTRAST_ENHANCEMENT);
}
/**
* Specified the supported Histogram Enhancement algorithms.
*
* @todo in the future this should be pluggable.
*/
private final static Set<String> SUPPORTED_HE_ALGORITHMS;
/**
* This are the different types f histogram equalization that we support for
* the moment. MOre should be added soon.
*
*/
static {
//load the contrast enhancement operations
final HashSet<String> heAlg = new HashSet<String>(2, 1.0f);
heAlg.add("NORMALIZE");
heAlg.add("HISTOGRAM");
heAlg.add("LOGARITHMIC");
heAlg.add("EXPONENTIAL");
SUPPORTED_HE_ALGORITHMS = Collections.unmodifiableSet(heAlg);
//load, if it is needed the GenericPiecewise tranformation
try{
new ParameterBlockJAI(GenericPiecewise.OPERATION_NAME);
}catch (final Exception e) {
GenericPiecewise.register(JAI.getDefaultInstance());
}
}
/** Histogram Enhancement algorithm to use. */
private String type = null;
/**
* Value we'll use for the gamma correction operation.
*/
private double gammaValue = Double.NaN;
/*
* (non-Javadoc)
*
* @see org.geotools.renderer.lite.gridcoverage2d.StyleVisitorAdapter#visit(org.geotools.styling.ContrastEnhancement)
*/
public void visit(final ContrastEnhancement ce) {
// /////////////////////////////////////////////////////////////////////
//
// Do nothing if we don't have a valid ContrastEnhancement element. This
// would protect us against bad SLDs
//
// /////////////////////////////////////////////////////////////////////
if (ce == null){
return;
}
// /////////////////////////////////////////////////////////////////////
//
// TYPE of the operation to perform
//
// /////////////////////////////////////////////////////////////////////
final Expression expType = ce.getType();
if (expType != null) {
final String type = expType.evaluate(null, String.class);
if (type != null) {
this.type = type.toUpperCase();
if (!SUPPORTED_HE_ALGORITHMS.contains(type.toUpperCase()))
throw new IllegalArgumentException(Errors.format(ErrorKeys.OPERATION_NOT_FOUND_$1, type.toUpperCase()));
}
}
// /////////////////////////////////////////////////////////////////////
//
// GAMMA
//
// /////////////////////////////////////////////////////////////////////
final Expression gamma = ce.getGammaValue();
if (gamma != null) {
final Number number = gamma.evaluate(null, Double.class);
if (number != null) {
gammaValue = number.doubleValue();
// check the gamma value
if (gammaValue < 0)
throw new IllegalArgumentException(Errors.format(ErrorKeys.ILLEGAL_ARGUMENT_$2, "Gamma", number));
if (Double.isNaN(gammaValue) || Double.isInfinite(gammaValue))
throw new IllegalArgumentException(Errors.format(ErrorKeys.ILLEGAL_ARGUMENT_$2, "Gamma", number));
}
}
}
/**
* Default constructor
*/
public ContrastEnhancementNode() {
this(null);
}
/**
* Constructor for a {@link ContrastEnhancementNode} which allows to specify
* a {@link Hints} instance to control internal factory machinery.
*
* @param hints
* {@link Hints} instance to control internal factory machinery.
*/
public ContrastEnhancementNode(final Hints hints) {
super(
1,
hints,
SimpleInternationalString.wrap("ContrastEnhancementNode"),
SimpleInternationalString
.wrap("Node which applies ContrastEnhancement following SLD 1.0 spec."));
}
/*
* (non-Javadoc)
*
* @see org.geotools.renderer.lite.gridcoverage2d.StyleVisitorCoverageProcessingNodeAdapter#execute()
*/
@SuppressWarnings("unchecked")
protected GridCoverage2D execute() {
final Hints hints = getHints();
// /////////////////////////////////////////////////////////////////////
//
// Get the sources and see what we got to do. Note that if we have more
// than once source we'll use only the first one but we'll
//
// /////////////////////////////////////////////////////////////////////
final List<CoverageProcessingNode> sources = this.getSources();
if (sources != null && !sources.isEmpty()) {
final GridCoverage2D source = (GridCoverage2D) getSource(0).getOutput();
GridCoverageRendererUtilities.ensureSourceNotNull(source, this.getName().toString());
GridCoverage2D output;
if ((!Double.isNaN(gammaValue) &&
!Double.isInfinite(gammaValue) &&
!(Math.abs(gammaValue -1)<1E-6))||
(type != null && type.length() > 0)) {
// /////////////////////////////////////////////////////////////////////
//
// We have a valid gamma value, let's go ahead.
//
// /////////////////////////////////////////////////////////////////////
final RenderedImage sourceImage = source.getRenderedImage();
// /////////////////////////////////////////////////////////////////////
//
// PREPARATION
//
// /////////////////////////////////////////////////////////////////////
// //
//
// Get the source image and if necessary convert it to use a
// ComponentColorModel. This way we are sure we will have a
// visible image most part of the time.
//
// //
////
//
// @todo TODO HACK we need to convert to byte the image when going to
// apply HISTOGRAM anyway
//
////
RenderedImage initialImage;
if(type!=null&&type.equalsIgnoreCase("HISTOGRAM"))
{
initialImage =
new ImageWorker(sourceImage)
.setRenderingHints(hints)
.forceComponentColorModel()
.rescaleToBytes()
.getRenderedImage();
}
else
{
initialImage =
new ImageWorker(sourceImage)
.setRenderingHints(hints)
.forceComponentColorModel()
.getRenderedImage();
}
final int numbands = initialImage.getSampleModel().getNumBands();
// //
//
// Save the alpha band if present, in order to put it back
// later in the loop. We are not going to use it anyway for
// the IHS conversion.
//
// //
RenderedImage alphaBand = null;
if (numbands % 2 == 0) {
// get the alpha band
alphaBand = new ImageWorker(initialImage)
.setRenderingHints(hints).retainLastBand()
.getRenderedImage();
// strip the alpha band from the original image
initialImage = new ImageWorker(initialImage)
.setRenderingHints(hints).retainBands(numbands - 1)
.getRenderedImage();
}
// //
//
// Get the single band to work on, which might be the
// intensity for RGB(A) or the GRAY channel for Gray(A)
//
// //
RenderedImage intensityImage = null;
RenderedImage hChannel = null;
RenderedImage sChannel = null;
final boolean intensity;
RenderedImage IHS = null;
if (numbands > 1) {
// convert the prepared image to IHS colorspace to work
// on I band
IHS = new ImageWorker(initialImage)
.setRenderingHints(hints).forceColorSpaceIHS()
.getRenderedImage();
// get the various singular bands
intensityImage = new ImageWorker(IHS).setRenderingHints(
hints).retainFirstBand().getRenderedImage();
sChannel = new ImageWorker(IHS).setRenderingHints(hints)
.retainLastBand().getRenderedImage();
hChannel = new ImageWorker(IHS).setRenderingHints(hints)
.retainBands(new int[] { 1 }).getRenderedImage();
intensity = true;
} else {
// //
//
// we have only one band we don't need to go to IHS
//
// //
intensityImage = initialImage;
intensity = false;
}
// /////////////////////////////////////////////////////////////////////
//
// HISTOGRAM ENHANCEMENT
//
//
//
// /////////////////////////////////////////////////////////////////////
final RenderedImage heImage = performContrastEnhancement(intensityImage, hints);
// /////////////////////////////////////////////////////////////////////
//
// GAMMA CORRECTION
//
// Lookup for building the actual lut that caches the gamma
// correction function's values.
//
// /////////////////////////////////////////////////////////////////////
RenderedImage finalImage = performGammaCorrection(heImage,hints);
// /////////////////////////////////////////////////////////////////////
//
// POSTPROCESSING
//
// Take care of the intermediated image we left back. This
// means, handle the fact that we might have gone to IHS and
// the alpha band.
//
// /////////////////////////////////////////////////////////////////////
if (intensity) {
// //
//
// IHS --> RGB
//
// Let's merge the modified IHS image. The message on
// the mailing list (see comments for this class)
// mentioned that it is required to pass a RenderingHing
// with a ImageLayout with the IHS color
// model.
//
// //
final ImageLayout imageLayout = new ImageLayout();
imageLayout.setColorModel(IHS.getColorModel());
imageLayout.setSampleModel(IHS.getSampleModel());
final RenderingHints rendHints = new RenderingHints(Collections.EMPTY_MAP);
rendHints.add(hints);
rendHints.add(new RenderingHints(JAI.KEY_IMAGE_LAYOUT,imageLayout));
// merge and go to rgb again
final ParameterBlock pb = new ParameterBlock();
pb.addSource(finalImage);
pb.addSource(hChannel);
pb.addSource(sChannel);
finalImage = JAI.create("bandmerge", pb, rendHints);
finalImage = new ImageWorker(finalImage).setRenderingHints(hints).forceColorSpaceRGB().getRenderedImage();
}
// //
//
// ALPHA BAND
//
// Let's merge the alpha band with the image we have rebuilt.
//
// //
if (alphaBand != null) {
final ColorModel cm = new ComponentColorModel(
numbands >= 3 ? ColorSpace
.getInstance(ColorSpace.CS_sRGB)
: ColorSpace
.getInstance(ColorSpace.CS_GRAY),
numbands >= 3 ? new int[] { 8, 8, 8, 8 }
: new int[] { 8, 8 }, true, false,
Transparency.TRANSLUCENT, DataBuffer.TYPE_BYTE);
final ImageLayout imageLayout = new ImageLayout();
imageLayout.setColorModel(cm);
imageLayout.setSampleModel(cm.createCompatibleSampleModel(finalImage.getWidth(), finalImage.getHeight()));
// merge and go to rgb
finalImage =
new ImageWorker(finalImage)
.setRenderingHints(hints)
.setRenderingHint(JAI.KEY_IMAGE_LAYOUT,imageLayout)
.addBand(alphaBand, false)
.getRenderedOperation();
}
// /////////////////////////////////////////////////////////////////////
//
// OUTPUT
//
// /////////////////////////////////////////////////////////////////////
final int numSourceBands=source.getNumSampleDimensions();
final int numActualBands= finalImage.getSampleModel().getNumBands();
final GridCoverageFactory factory = getCoverageFactory();
final HashMap<Object,Object> props = new HashMap<Object,Object>();
if(source.getProperties() != null) {
props.putAll(source.getProperties());
}
if(numActualBands==numSourceBands) {
final String name = "ce_coverage" + source.getName();
output = factory.create(
name,
finalImage,
(GridGeometry2D)source.getGridGeometry(),
source.getSampleDimensions(),
new GridCoverage[]{source},
props);
} else {
// replicate input bands
final GridSampleDimension sd[]= new GridSampleDimension[numActualBands];
for(int i=0;i<numActualBands;i++)
sd[i]=(GridSampleDimension) source.getSampleDimension(0);
output = factory.create(
"ce_coverage"+source.getName().toString(),
finalImage,
(GridGeometry2D)source.getGridGeometry(),
sd,
new GridCoverage[]{source},
props);
}
} else
// /////////////////////////////////////////////////////////////////////
//
// We do not have a valid gamma value, let's try with a
// conservative approach that is, let's forward the source
// coverage intact.
//
// /////////////////////////////////////////////////////////////////////
output = source;
return output;
}
throw new IllegalStateException(Errors.format(
ErrorKeys.SOURCE_CANT_BE_NULL_$1, this.getName().toString()));
}
/**
* Performs a contrast enhancement operation on the input image. Note that
* not all the contrast enhancement operations have been implemented in a
* way that is generic enough o handle all data types.
*
* @param inputImage the input {@link RenderedImage} to work on.
* @param hints {@link Hints} to control the contrast enhancement process.
* @return a {@link RenderedImage} on which a contrast enhancement has been performed.
*/
private RenderedImage performContrastEnhancement(
RenderedImage inputImage,
final Hints hints) {
if (type != null && type.length() > 0) {
assert inputImage.getSampleModel().getNumBands() == 1:inputImage;
final int dataType=inputImage.getSampleModel().getDataType();
// /////////////////////////////////////////////////////////////////////
//
// Histogram Normalization
//
//
//
// /////////////////////////////////////////////////////////////////////
if (type.equalsIgnoreCase("NORMALIZE")) {
//step 1 do the extrema to get the statistics for this image
final RenderedOp image = ExtremaDescriptor.create(inputImage,
null, Integer.valueOf(1), Integer.valueOf(1), null,
Integer.valueOf(1), null);
final double[][] extrema = (double[][]) image
.getProperty("extrema");
final int numBands = extrema[0].length;
assert numBands == 1:image;
// //
//
// Shortcut fr byte datatype
//
// //
if(dataType==DataBuffer.TYPE_BYTE){
////
//
// Optimisation for byte images, we use the lookup operation
//
////
if (extrema[1][0] == 255 && extrema[0][0] == 0)
return inputImage;
final double delta = extrema[1][0] - extrema[0][0];
final double scale = 255 / delta;
final double offset = -scale * extrema[0][0];
//create the lookup table
final byte[] lut = new byte[256];
for (int i = 1; i < lut.length; i++)
lut[i] = (byte) (scale * i + offset + 0.5d);
//do the actual lookup
final LookupTableJAI lookup = new LookupTableJAI(lut);
final ParameterBlock pb = new ParameterBlock();
pb.addSource(inputImage);
pb.add(lookup);
return JAI.create("lookup", pb, hints);
}
////
//
// General case, we use the rescale in order to stretch the values to highest and lowest dim
//
////
//get the correct dim for this data type
final double maximum=ColorUtilities.getMaximum(dataType);
final double minimum=ColorUtilities.getMinimum(dataType);
if (extrema[1][0] == maximum && extrema[0][0] == minimum)
return inputImage;
//compute the scale factors
final double delta = extrema[1][0] - extrema[0][0];
final double scale = (maximum -minimum)/ delta;
final double offset = minimum - scale * extrema[0][0];
//do the actual rescale
final ParameterBlock pb = new ParameterBlock();
pb.addSource(inputImage);
pb.add(new double []{scale});
pb.add(new double []{offset});
return JAI.create("rescale", pb, hints);
}
// /////////////////////////////////////////////////////////////////////
//
// EXPONENTIAL Normalization
//
//
//
// /////////////////////////////////////////////////////////////////////
if (type.equalsIgnoreCase("EXPONENTIAL")) {
if(dataType==DataBuffer.TYPE_BYTE){
////
//
// Optimisation for byte images
//
////
final byte lut[] = new byte[256];
final double normalizationFactor=255.0;
final double correctionFactor=255.0/(Math.E-1);
for (int i = 1; i < lut.length; i++)
lut[i] = (byte) (0.5f + correctionFactor * (Math.exp(i / normalizationFactor) - 1.0));
return LookupDescriptor.create(inputImage,
new LookupTableJAI(lut), hints);
}
////
//
// General case, we use the piecewise1D transform
//
////
//
// STEP 1 do the extrema
//
////
//step 1 do the extrema to get the statistics for this image
final RenderedOp statistics = ExtremaDescriptor.create(inputImage,
null, Integer.valueOf(1), Integer.valueOf(1), null,
Integer.valueOf(1), null);
final double[] minimum=(double[]) statistics.getProperty("minimum");
final double[] maximum=(double[]) statistics.getProperty("maximum");
final double normalizationFactor=maximum[0];
final double correctionFactor=normalizationFactor/(Math.E-1);
////
//
// STEP 2 do the gamma correction by using generic piecewise
//
////
final DefaultPiecewiseTransform1DElement mainElement = DefaultPiecewiseTransform1DElement.create(
"exponential-contrast-enhancement-transform", NumberRange.create(minimum[0],maximum[0]),
new MathTransform1DAdapter() {
/*
* (non-Javadoc)
* @see org.opengis.referencing.operation.MathTransform1D#derivative(double)
*/
public double derivative(double value)
throws TransformException {
throw new UnsupportedOperationException(Errors.format(ErrorKeys.UNSUPPORTED_OPERATION_$1));
}
public boolean isIdentity() {
return false;
}
/*
* (non-Javadoc)
* @see org.opengis.referencing.operation.MathTransform1D#transform(double)
*/
public double transform(double value)
throws TransformException {
value = correctionFactor*(Math.exp(value/normalizationFactor)-1);
return value;
}
});
final PiecewiseTransform1D<DefaultPiecewiseTransform1DElement> transform = new DefaultPiecewiseTransform1D<DefaultPiecewiseTransform1DElement> (
new DefaultPiecewiseTransform1DElement[] {mainElement},0);
final ParameterBlockJAI pbj = new ParameterBlockJAI(
GenericPiecewise.OPERATION_NAME);
pbj.addSource(inputImage);
pbj.setParameter("Domain1D", transform);
pbj.setParameter("bandIndex", Integer.valueOf(0));
return JAI.create(
GenericPiecewise.OPERATION_NAME, pbj);
}
if (type.equalsIgnoreCase("LOGARITHMIC")) {
// /////////////////////////////////////////////////////////////////////
//
// Logarithm Normalization
//
//
//
// /////////////////////////////////////////////////////////////////////
if(dataType==DataBuffer.TYPE_BYTE){
////
//
// Optimisation for byte images m we use lookup
//
////
final byte lut[] = new byte[256];
final double normalizationFactor=255.0;
final double correctionFactor=100.0;
for (int i = 1; i < lut.length; i++)
lut[i] = (byte) (0.5f + normalizationFactor * Math.log((i * correctionFactor / normalizationFactor+ 1.0)));
return LookupDescriptor.create(inputImage,
new LookupTableJAI(lut), hints);
}
////
//
// General case
//
////
//define a specific piecewise for the logarithm
////
//
// STEP 1 do the extrema
//
////
//step 1 do the extrema to get the statistics for this image
final RenderedOp statistics = ExtremaDescriptor.create(inputImage,
null, Integer.valueOf(1), Integer.valueOf(1), null,
Integer.valueOf(1), null);
final double[] minimum=(double[]) statistics.getProperty("minimum");
final double[] maximum=(double[]) statistics.getProperty("maximum");
final double normalizationFactor=maximum[0];
final double correctionFactor=100.0;
////
//
// STEP 2 do the gamma correction by using generic piecewise
//
////
final DefaultPiecewiseTransform1DElement mainElement = DefaultPiecewiseTransform1DElement.create(
"logarithmic-contrast-enhancement-transform", NumberRange.create(minimum[0],maximum[0]),
new MathTransform1DAdapter() {
/*
* (non-Javadoc)
* @see org.opengis.referencing.operation.MathTransform1D#derivative(double)
*/
public double derivative(double value)
throws TransformException {
throw new UnsupportedOperationException(Errors.format(ErrorKeys.UNSUPPORTED_OPERATION_$1));
}
public boolean isIdentity() {
return false;
}
/*
* (non-Javadoc)
* @see org.opengis.referencing.operation.MathTransform1D#transform(double)
*/
public double transform(double value)
throws TransformException {
value =normalizationFactor*Math.log(1+(value*correctionFactor/normalizationFactor));
return value;
}
});
final PiecewiseTransform1D<DefaultPiecewiseTransform1DElement> transform = new DefaultPiecewiseTransform1D<DefaultPiecewiseTransform1DElement> (
new DefaultPiecewiseTransform1DElement[] {mainElement},0);
final ParameterBlockJAI pbj = new ParameterBlockJAI(
GenericPiecewise.OPERATION_NAME);
pbj.addSource(inputImage);
pbj.setParameter("Domain1D", transform);
pbj.setParameter("bandIndex", Integer.valueOf(0));
return JAI.create(
GenericPiecewise.OPERATION_NAME, pbj);
}
if (type.equalsIgnoreCase("HISTOGRAM")) {
// /////////////////////////////////////////////////////////////////////
//
// Histogram Equalization
//
// IT WORKS ONLY ON BYTE DATA TYPE!!!
//
// /////////////////////////////////////////////////////////////////////
//convert the input image to 8 bit
inputImage=new ImageWorker(inputImage).rescaleToBytes().getRenderedImage();
// compute the histogram
final RenderedOp hist = HistogramDescriptor.create(inputImage,
null, Integer.valueOf(1), Integer.valueOf(1),
new int[] { 256 }, new double[] { 0 },
new double[] { 256 }, null);
final Histogram h = (Histogram) hist.getProperty("histogram");
// now compute the PDF and the CDF for the original image
final byte[] cumulative = new byte[h.getNumBins(0)];
// sum of bins (we might have excluded 0 hence we cannot really
// optimise)
float totalBinSum = 0;
for (int i = 0; i < cumulative.length; i++) {
totalBinSum += h.getBinSize(0, i);
}
// this is the scale factor for the histogram equalization
// process
final float scale = (float) (h.getHighValue(0) - 1 - h.getLowValue(0))/ totalBinSum;
float sum = 0;
for (int i = 1; i < cumulative.length; i++) {
sum += h.getBinSize(0, i - 1);
cumulative[i] = (byte) ((sum * scale + h.getLowValue(0)) + .5F);
}
final LookupTableJAI lookup = new LookupTableJAI(cumulative);
final ParameterBlock pb = new ParameterBlock();
pb.addSource(hist);
pb.add(lookup);
return JAI.create("lookup", pb, hints);
}
}
return inputImage;
}
/**
* Performs a gamma correction operation on the input image.
*
* @param inputImage the input {@link RenderedImage} to work on.
* @param hints {@link Hints} to control the contrast enhancement process.
* @return a {@link RenderedImage} on which a gamma correction has been performed.
*/
private RenderedImage performGammaCorrection(
final RenderedImage inputImage,
final Hints hints) {
//note that we should work on a single band
assert inputImage.getSampleModel().getNumBands() == 1:inputImage;
final int dataType=inputImage.getSampleModel().getDataType();
RenderedImage result=inputImage;
if (!Double.isNaN(gammaValue) && Math.abs(gammaValue - 1.0) > 1E-6) {
if (dataType == DataBuffer.TYPE_BYTE) {
////
//
// Byte case, use lookup to optimize
//
////
final byte[] lut = new byte[256];
for (int i = 1; i < lut.length; i++)
lut[i] = (byte) (255.0 * Math.pow(i / 255.0, gammaValue) + 0.5d);
// apply the operation now
final LookupTableJAI lookup = new LookupTableJAI(lut);
final ParameterBlock pb = new ParameterBlock();
pb.addSource(inputImage);
pb.add(lookup);
result = JAI.create("lookup", pb, hints);
}
else
{
////
//
// Generic case
//
////
//
// STEP 1 do the extrema
//
////
//step 1 do the extrema to get the statistics for this image
final RenderedOp statistics = ExtremaDescriptor.create(inputImage,
null, Integer.valueOf(1), Integer.valueOf(1), null,
Integer.valueOf(1), null);
final double[] minimum=(double[]) statistics.getProperty("minimum");
final double[] maximum=(double[]) statistics.getProperty("maximum");
final double scale = (maximum[0]-minimum[0])/(MAX_VALUE-MIN_VALUE);
final double offset = minimum[0] - MIN_VALUE*scale;
////
//
// STEP 2 do the gamma correction by using generic piecewise
//
////
final DefaultPiecewiseTransform1DElement mainElement = DefaultPiecewiseTransform1DElement.create(
"gamma-correction-transform", NumberRange.create(minimum[0],maximum[0]),
new MathTransform1DAdapter() {
/*
* (non-Javadoc)
* @see org.opengis.referencing.operation.MathTransform1D#derivative(double)
*/
public double derivative(double value)
throws TransformException {
throw new UnsupportedOperationException(Errors.format(ErrorKeys.UNSUPPORTED_OPERATION_$1));
}
public boolean isIdentity() {
return false;
}
/*
* (non-Javadoc)
* @see org.opengis.referencing.operation.MathTransform1D#transform(double)
*/
public double transform(double value)
throws TransformException {
value = (value-offset)/scale;
return offset+Math.pow(value, gammaValue)*scale;
}
});
final PiecewiseTransform1D<DefaultPiecewiseTransform1DElement> transform = new DefaultPiecewiseTransform1D<DefaultPiecewiseTransform1DElement> (
new DefaultPiecewiseTransform1DElement[] {mainElement},0);
final ParameterBlockJAI pbj = new ParameterBlockJAI(
GenericPiecewise.OPERATION_NAME);
pbj.addSource(inputImage);
pbj.setParameter("Domain1D", transform);
pbj.setParameter("bandIndex", Integer.valueOf(0));
result = JAI.create(
GenericPiecewise.OPERATION_NAME, pbj);
}
}
assert result.getSampleModel().getNumBands() == 1:result;
return result;
}
}