/*
* This file is part of the LIRE project: http://lire-project.net
* LIRE is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* LIRE 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 General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with LIRE; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*
* We kindly ask you to refer the any or one of the following publications in
* any publication mentioning or employing Lire:
*
* Lux Mathias, Savvas A. Chatzichristofis. Lire: Lucene Image Retrieval –
* An Extensible Java CBIR Library. In proceedings of the 16th ACM International
* Conference on Multimedia, pp. 1085-1088, Vancouver, Canada, 2008
* URL: http://doi.acm.org/10.1145/1459359.1459577
*
* Lux Mathias. Content Based Image Retrieval with LIRE. In proceedings of the
* 19th ACM International Conference on Multimedia, pp. 735-738, Scottsdale,
* Arizona, USA, 2011
* URL: http://dl.acm.org/citation.cfm?id=2072432
*
* Mathias Lux, Oge Marques. Visual Information Retrieval using Java and LIRE
* Morgan & Claypool, 2013
* URL: http://www.morganclaypool.com/doi/abs/10.2200/S00468ED1V01Y201301ICR025
*
* Copyright statement:
* --------------------
* (c) 2002-2013 by Mathias Lux (mathias@juggle.at)
* http://www.semanticmetadata.net/lire, http://www.lire-project.net
*/
package net.semanticmetadata.lire.utils.cv;
import net.semanticmetadata.lire.imageanalysis.filters.FastBilateralFilter;
import net.semanticmetadata.lire.imageanalysis.filters.IndexedIntArray;
import net.semanticmetadata.lire.utils.FileUtils;
import net.semanticmetadata.lire.utils.MetricsUtils;
import javax.imageio.ImageIO;
import java.awt.image.BufferedImage;
import java.awt.image.WritableRaster;
import java.io.File;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;
/**
* A simple pixel clustering algorithm based on RGB, k-means and L2 distance.
* @author Mathias Lux, mathias@juggle.at - 20.09.13 10:39
*/
public class PixelClustering {
private static int numberOfColors = 32;
public static BufferedImage clusterPixels(BufferedImage img) {
// Apply Bilateral Filtering before actually classifying the pixels ...
BufferedImage b = new BufferedImage(img.getWidth(), img.getHeight(), BufferedImage.TYPE_INT_RGB);
b.getGraphics().drawImage(img, 0, 0, null);
IndexedIntArray src = new IndexedIntArray(new int[img.getWidth()*img.getHeight()], 0);
IndexedIntArray dst = new IndexedIntArray(new int[img.getWidth()*img.getHeight()], 0);
b.getRaster().getDataElements(0, 0, img.getWidth(), img.getHeight(), src.array);
FastBilateralFilter fbf = new FastBilateralFilter(img.getWidth(), img.getHeight(), img.getWidth(), 90f, 0.3f);
fbf.apply(src, dst);
b.getRaster().setDataElements(0, 0, img.getWidth(), img.getHeight(), dst.array);
/*
try {
ImageIO.write(b, "png", new File("out_filtered.png"));
ImageIO.write(img, "png", new File("out_original.png"));
} catch (IOException e) {
e.printStackTrace(); //To change body of catch statement use File | Settings | File Templates.
}
*/
img.getGraphics().drawImage(b, 0, 0, null);
WritableRaster r = img.getRaster();
// quantize image colors with k-means:
ArrayList<double[]> pixels = new ArrayList<double[]>(r.getHeight()*r.getWidth());
for (int x = 0; x < r.getWidth(); x++) {
for (int y = 0; y < r.getHeight(); y++) {
double[] tmpPixel = new double[3];
// double[] tmpPixel = new double[5]; // use this one if you want connected patches.
r.getPixel(x, y, tmpPixel);
assert(tmpPixel[0]<256);
assert(tmpPixel[1]<256);
assert(tmpPixel[2]<256);
// tmpPixel[3] = x*255f/(double)r.getWidth(); // use this one if you want connected patches.
// tmpPixel[4] = y*255f/(double)r.getHeight();
pixels.add(tmpPixel);
}
}
// do the k-means
KMeans km = new KMeans(pixels, numberOfColors);
for (int i=0; i<15; i++)
km.step();
List<double[]> means = km.getMeans();
double[] tmpPixel = new double[3];
// double[] tmpPixel = new double[5]; // use this one if you want connected patches.
for (int x = 0; x < r.getWidth(); x++) {
for (int y = 0; y < r.getHeight(); y++) {
r.getPixel(x, y, tmpPixel);
// tmpPixel[3] = x*255f/(double)r.getWidth(); // use this one if you want connected patches.
// tmpPixel[4] = y*255f/(double)r.getHeight();
int num = -1;
int count = 0;
double distance=-1, tmpDistance=-1;
for (Iterator<double[]> iterator = means.iterator(); iterator.hasNext(); ) {
double[] next = iterator.next();
distance = MetricsUtils.distL2(next, tmpPixel);
if (num < 0 || distance < tmpDistance) {
num = count;
tmpDistance = distance;
}
count++;
}
tmpPixel[0] = Math.floor(means.get(num)[0]);
tmpPixel[1] = Math.floor(means.get(num)[1]);
tmpPixel[2] = Math.floor(means.get(num)[2]);
r.setPixel(x,y,tmpPixel);
}
}
return img;
}
public static void main(String[] args) throws IOException {
ArrayList<File> files = FileUtils.getAllImageFiles(new File("D:\\Temp\\tmp"), false);
int count = 10;
for (Iterator<File> iterator = files.iterator(); iterator.hasNext(); ) {
File next = iterator.next();
BufferedImage img = ImageIO.read(next);
BufferedImage toWrite = new BufferedImage(img.getWidth()*2, img.getHeight(), BufferedImage.TYPE_INT_RGB);
toWrite.getGraphics().drawImage(img, 0, 0, null);
BufferedImage bufferedImage = clusterPixels(img);
toWrite.getGraphics().drawImage(bufferedImage, img.getWidth(), 0, null);
ImageIO.write(toWrite, "png", new File("out_test_"+count+".png"));
count++;
}
}
}