net.semanticmetadata.lire.indexers.hashing.LocalitySensitiveHashing Maven / Gradle / Ivy
The newest version!
/*
* 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 ([email protected])
* http://www.semanticmetadata.net/lire, http://www.lire-project.net
*
* Updated: 02.06.13 10:27
*/
package net.semanticmetadata.lire.indexers.hashing;
import java.io.*;
import java.util.zip.GZIPInputStream;
import java.util.zip.GZIPOutputStream;
/**
* Each feature vector v with dimension d gets k hashes from a hash bundle h(v) = (h^1(v), h^2(v), ..., h^k(v)) with
* h^i(v) = (a^i*v + b^i)/w (rounded down), with a^i from R^d and b^i in [0,w)
* If m of the k hashes match, then we assume that the feature vectors belong to similar images. Note that m*k has to be bigger than d!
* If a^i is drawn from a normal (Gaussian) distribution LSH approximates L2.
*
* Note that this is just to be used with bounded (normalized) descriptors.
*
* @author Mathias Lux, [email protected]
* Created: 04.06.12, 13:42
*/
public class LocalitySensitiveHashing {
private static String name = "lshHashFunctions.obj";
private static int dimensions = 250; // max d
public static int numFunctionBundles = 50; // k
public static double binLength = 10; // w
private static double[][] hashA = null; // a
private static double[] hashB = null; // b
private static double dilation = 1d; // defines how "stretched out" the hash values are.
/**
* Writes a new file to disk to be read for hashing with LSH.
*
* @throws java.io.IOException
*/
public static void generateHashFunctions() throws IOException {
File hashFile = new File(name);
if (!hashFile.exists()) {
ObjectOutputStream oos = new ObjectOutputStream(new GZIPOutputStream(new FileOutputStream(hashFile)));
oos.writeInt(dimensions);
oos.writeInt(numFunctionBundles);
for (int c = 0; c < numFunctionBundles; c++) {
oos.writeFloat((float) (Math.random() * binLength));
}
for (int c = 0; c < numFunctionBundles; c++) {
for (int j = 0; j < dimensions; j++) {
oos.writeFloat((float) (drawNumber() * dilation));
}
}
oos.close();
} else {
System.err.println("Hashes could not be written: " + name + " already exists");
}
}
public static void generateHashFunctions(String name) throws IOException {
File hashFile = new File(name);
if (!hashFile.exists()) {
ObjectOutputStream oos = new ObjectOutputStream(new GZIPOutputStream(new FileOutputStream(hashFile)));
oos.writeInt(dimensions);
oos.writeInt(numFunctionBundles);
for (int c = 0; c < numFunctionBundles; c++) {
oos.writeFloat((float) (Math.random() * binLength));
}
for (int c = 0; c < numFunctionBundles; c++) {
for (int j = 0; j < dimensions; j++) {
oos.writeFloat((float) (drawNumber() * dilation));
}
}
oos.close();
} else {
System.err.println("Hashes could not be written: " + name + " already exists");
}
}
/**
* Reads a file from disk and sets the hash functions.
*
* @return
* @throws IOException
* @see LocalitySensitiveHashing#generateHashFunctions()
*/
public static double[][] readHashFunctions() throws IOException {
return readHashFunctions(new FileInputStream(name));
}
public static double[][] readHashFunctions(InputStream in) throws IOException {
ObjectInputStream ois = new ObjectInputStream(new GZIPInputStream(in));
dimensions = ois.readInt();
numFunctionBundles = ois.readInt();
double[] tmpB = new double[numFunctionBundles];
for (int k = 0; k < numFunctionBundles; k++) {
tmpB[k] = ois.readFloat();
}
LocalitySensitiveHashing.hashB = tmpB;
double[][] hashFunctions = new double[numFunctionBundles][dimensions];
for (int i = 0; i < hashFunctions.length; i++) {
double[] functionBundle = hashFunctions[i];
for (int j = 0; j < functionBundle.length; j++) {
functionBundle[j] = ois.readFloat();
}
}
LocalitySensitiveHashing.hashA = hashFunctions;
return hashFunctions;
}
/**
* Generates the hashes from the given hash bundles.
*
* @param histogram
* @return
*/
public static int[] generateHashes(double[] histogram) {
double product;
int[] result = new int[numFunctionBundles];
for (int k = 0; k < numFunctionBundles; k++) {
product = 0;
for (int i = 0; i < histogram.length; i++) {
product += histogram[i] * hashA[k][i];
}
result[k] = (int) Math.floor((product + hashB[k]) / binLength);
}
return result;
}
/**
* Returns a random number distributed with standard normal distribution based on the Box-Muller method.
*
* @return
*/
private static double drawNumber() {
double u, v, s;
do {
u = Math.random() * 2 - 1;
v = Math.random() * 2 - 1;
s = u * u + v * v;
} while (s == 0 || s >= 1);
return u * Math.sqrt(-2d * Math.log(s) / s);
// return Math.sqrt(-2d * Math.log(Math.random())) * Math.cos(2d * Math.PI * Math.random());
}
public static void main(String[] args) {
try {
generateHashFunctions();
} catch (IOException e) {
e.printStackTrace();
}
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy