net.semanticmetadata.lire.indexers.hashing

类 MetricSpaces

java.lang.Object

net.semanticmetadata.lire.indexers.hashing.MetricSpaces

public class MetricSpaces
extends java.lang.Object

This hashing approach implements the proximity approach "metric spaces" based on the work of Giuseppe Amato. See Gennaro, Claudio, et al. "An approach to content-based image retrieval based on the Lucene search engine library." Research and Advanced Technology for Digital Libraries. Springer Berlin Heidelberg, 2010. 55-66.

While it's actually not a hashing routine it's an easy way to use it within LIRE. Based on a pre-trained set of representatives, the features of the representatives are stored in a text file. For indexing and search they can be used to create hash like text snippets, that are then used to search & index.

作者:

Mathias Lux, mathias@juggle.at, 24.08.2015.

嵌套类概要

嵌套类

限定符和类型	类和说明
static class	MetricSpaces.Parameters For storing the parameters per feature.
static class	MetricSpaces.Result

构造器概要

构造器

构造器和说明
MetricSpaces()

方法概要

限定符和类型	方法和说明
static java.lang.String	generateBoostedQuery(GlobalFeature feature, int queryLength) Creates a text String to be used for search based on the reference points using term boosting instead of repetition.
static java.lang.String	generateHashString(GlobalFeature feature) Convenience method for generateHashString(GlobalFeature, int).
static java.lang.String	generateHashString(GlobalFeature feature, int queryLength) Creates a text String to be used for indexing and search based on the reference points.
static void	indexReferencePoints(java.lang.Class globalFeatureClass, int numberOfReferencePoints, int lenghtOfPostingList, java.io.File inFile, java.io.File outFile) Index reference points for use with a specific feature.
static MetricSpaces.Parameters	loadReferencePoints(java.io.File referencePoints) Init with a single file.
static void	main(java.lang.String[] args)
static boolean	supportsFeature(GlobalFeature feature)

从类继承的方法 java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

构造器详细资料

MetricSpaces

public MetricSpaces()

方法详细资料

main

public static void main(java.lang.String[] args)

indexReferencePoints

public static void indexReferencePoints(java.lang.Class globalFeatureClass,
                                        int numberOfReferencePoints,
                                        int lenghtOfPostingList,
                                        java.io.File inFile,
                                        java.io.File outFile)
                                 throws java.io.IOException,
                                        java.lang.IllegalAccessException,
                                        java.lang.InstantiationException

Index reference points for use with a specific feature. A file (infile) containing one image path per line is read and numberOfReferencePoints are randomly selected to serve for indexing and search. The resulting data points plus configuration are read written to a file (outfile).

参数:

globalFeatureClass - The feature class to be used, eg. CEDD, PHOG, etc.

numberOfReferencePoints - the number of reference points, eg. 5000

lenghtOfPostingList - the length of the posting list, ie. how many reference points per image are stored.

inFile - the file containing the image data, one image path per line

outFile - the output of processing.

抛出:

java.io.IOException

java.lang.IllegalAccessException

java.lang.InstantiationException

loadReferencePoints

public static MetricSpaces.Parameters loadReferencePoints(java.io.File referencePoints)
                                                   throws java.io.IOException,
                                                          java.lang.ClassNotFoundException,
                                                          java.lang.IllegalAccessException,
                                                          java.lang.InstantiationException

Init with a single file. In this file the feature class and all parameters are given. To create such a file see indexReferencePoints(Class, int, int, File, File). Note that you can load multiple files, one for each feature. If you load more than one per feature class, they will be overwritten.

参数:

referencePoints - is the outFile from the method indexReferencePoints(Class, int, int, File, File)

抛出:

java.io.IOException

java.lang.ClassNotFoundException

java.lang.IllegalAccessException

java.lang.InstantiationException

supportsFeature

public static boolean supportsFeature(GlobalFeature feature)

generateHashString

public static java.lang.String generateHashString(GlobalFeature feature)

Convenience method for generateHashString(GlobalFeature, int).

参数:

feature - feature the feature instance the string is generated for.

返回:

the text for the Lucene index.

generateHashString

public static java.lang.String generateHashString(GlobalFeature feature,
                                                  int queryLength)

Creates a text String to be used for indexing and search based on the reference points.

参数:

feature - the feature instance the string is generated for.

queryLength - the length of the posting list. If 0, then it's set to the preset.

返回:

the text for the Lucene index.

generateBoostedQuery

public static java.lang.String generateBoostedQuery(GlobalFeature feature,
                                                    int queryLength)

Creates a text String to be used for search based on the reference points using term boosting instead of repetition.

参数:

feature - the feature instance the string is generated for.

queryLength - the length of the posting list. If 0, then it's set to the preset.

返回:

the text for the Lucene index.