org.openimaj.knn.pq.FloatSDCNearestNeighbours Maven / Gradle / Ivy

Go to download
/*
	AUTOMATICALLY GENERATED BY jTemp FROM
	/Users/jsh2/Work/openimaj/target/checkout/machine-learning/nearest-neighbour/src/main/jtemp/org/openimaj/knn/pq/#T#SDCNearestNeighbours.jtemp
*/
/**
 * Copyright (c) 2011, The University of Southampton and the individual contributors.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modification,
 * are permitted provided that the following conditions are met:
 *
 *   * 	Redistributions of source code must retain the above copyright notice,
 * 	this list of conditions and the following disclaimer.
 *
 *   *	Redistributions in binary form must reproduce the above copyright notice,
 * 	this list of conditions and the following disclaimer in the documentation
 * 	and/or other materials provided with the distribution.
 *
 *   *	Neither the name of the University of Southampton nor the names of its
 * 	contributors may be used to endorse or promote products derived from this
 * 	software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */
 
 package org.openimaj.knn.pq;

import org.openimaj.citation.annotation.Reference;
import org.openimaj.citation.annotation.ReferenceType;
import org.openimaj.feature.FloatFVComparison;
import org.openimaj.util.pair.IntFloatPair;
import org.openimaj.util.queue.BoundedPriorityQueue;
/**
 * Nearest-neighbours using Symmetric Distance Computation (SDC) on Product
 * Quantised vectors. In SDC, both query and the database points are quantised.
 * Distances are calculated by look-up from precomputed tables of the distance
 * between all centroids for each subvector.
 * 
 * SDC has the same computational cost as ADC, but a higher error in the
 * computed distance, so its use is not recommended. This implementation is
 * provided for completeness only.
 * 
 * @author Jonathon Hare ([email protected])
 */
@Reference(
		type = ReferenceType.Article,
		author = { "Jegou, Herve", "Douze, Matthijs", "Schmid, Cordelia" },
		title = "Product Quantization for Nearest Neighbor Search",
		year = "2011",
		journal = "IEEE Trans. Pattern Anal. Mach. Intell.",
		pages = { "117", "", "128" },
		url = "http://dx.doi.org/10.1109/TPAMI.2010.57",
		month = "January",
		number = "1",
		publisher = "IEEE Computer Society",
		volume = "33",
		customData = {
				"issn", "0162-8828",
				"numpages", "12",
				"doi", "10.1109/TPAMI.2010.57",
				"acmid", "1916695",
				"address", "Washington, DC, USA",
				"keywords", "High-dimensional indexing, High-dimensional indexing, image indexing, very large databases, approximate search., approximate search., image indexing, very large databases"
		})
public class FloatSDCNearestNeighbours extends FloatADCNearestNeighbours {
	float[][][] distances;

	/**
	 * Construct the SDC with the given quantiser, centroids (corresponding to
	 * the quantiser's internal assigners), and data.
	 * 
	 * @param pq
	 *            the Product Quantiser
	 * @param pqCentroids
	 *            the centroids corresponding to the the Product Quantiser's
	 *            internal assigners.
	 * @param dataPoints
	 *            the data to index
	 */
	public FloatSDCNearestNeighbours(FloatProductQuantiser pq, float[][][] pqCentroids, float[][] dataPoints) {
		super(pq, dataPoints);

		this.distances = new float[pq.assigners.length][][];

		for (int i = 0; i < pq.assigners.length; i++) {
			final float[][] centroids = pqCentroids[i];

			distances[i] = new float[centroids.length][centroids.length];

			for (int j = 0; j < centroids.length; j++) {
				for (int k = j; k < centroids.length; k++) {
					distances[i][j][k] = (float) FloatFVComparison.SUM_SQUARE.compare(centroids[j], centroids[k]);
					distances[i][k][j] = distances[i][j][k];
				}
			}
		}
	}

	@Override
	protected void computeDistances(float[] fullQuery, BoundedPriorityQueue queue, IntFloatPair workingPair)
	{
		final byte[] query = pq.quantise(fullQuery);

		for (int i = 0; i < data.length; i++) {
			workingPair.first = i;
			workingPair.second = 0;

			for (int j = 0; j < query.length; j++) {
				workingPair.second += distances[j][query[j] + 128][data[i][j] + 128];
			}

			workingPair = queue.offerItem(workingPair);
		}
	}
}