Java DocumentBuilder类代码示例

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
/**
 * Index for each builder type
 * @param image
 * @param picture_id
 * @param prefix
 * @param builder
 * @param conf
 * @throws IOException
 */
private static void luceneIndexer(BufferedImage image, UUID picture_id, String prefix, DocumentBuilder builder, IndexWriterConfig conf)
    throws IOException
{
    File path = getPath(prefix);
    log.debug("creating indexed path " + path.getAbsolutePath());
    IndexWriter iw = new IndexWriter(FSDirectory.open(path), conf);

    try {
        Document document = builder.createDocument(image, picture_id.toString());
        iw.addDocument(document);

    } catch (Exception e) {
        System.err.println("Error reading image or indexing it.");
        e.printStackTrace();
    }

    // closing the IndexWriter
    iw.close();
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
@Override
public Field[] createDescriptorFields(BufferedImage image) {
    Field[] result = null;
    try {
        // extract features from image:
        List<Feature> features = extractor.computeSiftFeatures(image);
        result = new Field[features.size()];
        int count = 0;
        // create new document:
        for (Iterator<Feature> fit = features.iterator(); fit.hasNext(); ) {
            Feature f = fit.next();
            result[count] = new StoredField(DocumentBuilder.FIELD_NAME_SIFT, f.getByteArrayRepresentation());
            count++;
        }
    } catch (IOException e) {
        logger.severe(e.getMessage());
    }
    return result;
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public Document createDocument(BufferedImage image, String identifier) {
    Document doc = null;
    try {
        // extract features from image:
        List<Feature> features = extractor.computeSiftFeatures(image);
        // create new document:
        doc = new Document();
        for (Iterator<Feature> fit = features.iterator(); fit.hasNext(); ) {
            Feature f = fit.next();
            // add each feature to the document:
            doc.add(new StoredField(DocumentBuilder.FIELD_NAME_SIFT, f.getByteArrayRepresentation()));
        }
        if (identifier != null)
            doc.add(new StringField(DocumentBuilder.FIELD_NAME_IDENTIFIER, identifier, Field.Store.YES));
    } catch (IOException e) {
        logger.severe(e.getMessage());
    }
    return doc;
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
/**
     * Creates a document from a (non existent) one color image. Can be used for
     * color search.
     *
     * @param color the color for the image
     * @return the document for searching.
     */
    public static Document createColorOnlyDocument(Color color) {
        assert (color != null);
        // Create a one pixel image
        int imgWidth = 64;
        BufferedImage img = new BufferedImage(imgWidth, imgWidth, BufferedImage.TYPE_INT_RGB);
        Graphics2D g = (Graphics2D) img.getGraphics();
        g.setColor(color);
        g.fillRect(0, 0, imgWidth, imgWidth);
        // Hand it over to ScalableColor to create a descriptor:
        ColorLayout scd = new ColorLayout();
        scd.extract(img);
        // create the string representation
        String sc = scd.getStringRepresentation();
//        System.out.println("sc = " + sc);
//        System.out.println("color = " + color);
        logger.fine("Extraction from image finished");
        Document doc = new Document();
        if (sc != null)
            doc.add(new StringField(DocumentBuilder.FIELD_NAME_COLORLAYOUT, sc, Field.Store.YES));
        return doc;
    }
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
/**
 * Creates a fully fledged Document to be added to a Lucene index.
 *
 * @param image      the image to index. Cannot be NULL.
 * @param identifier an id for the image, for instance the filename or an URL. Can be NULL.
 * @return
 */
public Document createDocument(BufferedImage image, String identifier) {
    assert (image != null);

    // sangupta: create a new document else code below
    // will throw a NPE
    Document doc = new Document();

    if (identifier != null) {
        doc.add(new StringField(DocumentBuilder.FIELD_NAME_IDENTIFIER, identifier, Field.Store.YES));
    }

    Field[] fields = createDescriptorFields(image);
    for (int i = 0; i < fields.length; i++) {
        doc.add(fields[i]);
    }

    return doc;
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public Document createDocument(BufferedImage image, String identifier) throws FileNotFoundException {
    docsCreated = true;
    Document doc = new Document();
    if (identifier != null)
        doc.add(new StoredField(DocumentBuilder.FIELD_NAME_IDENTIFIER, identifier));
    if (builders.size() >= 1) {
        for (DocumentBuilder builder : builders) {
            Field[] fields = builder.createDescriptorFields(image);
            for (int i = 0; i < fields.length; i++) {
                Field field = fields[i];
                doc.add(field);
            }
        }
    }
    return doc;
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
private void indexFiles(ArrayList<String> images, DocumentBuilder builder, String indexPath) throws IOException {
        System.out.println(">> Indexing " + images.size() + " files.");
        IndexWriter iw = LuceneUtils.createIndexWriter(indexPath, true);
        int count = 0;
        long time = System.currentTimeMillis();
        for (String identifier : images) {
            Document doc = builder.createDocument(new FileInputStream(identifier), identifier);
            iw.addDocument(doc);
            count++;
            if (count % 100 == 0) System.out.println(count + " files indexed.");
//            if (count == 200) break;
        }
        long timeTaken = (System.currentTimeMillis() - time);
        float sec = ((float) timeTaken) / 1000f;

        System.out.println(sec + " seconds taken, " + (timeTaken / count) + " ms per image.");
        iw.close();
    }
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
private void testSearchSpeed(ArrayList<String> images, final Class featureClass) throws IOException {
    parallelIndexer = new ParallelIndexer(8, indexPath, testExtensive, true) {
        @Override
        public void addBuilders(ChainedDocumentBuilder builder) {
            builder.addBuilder(new GenericDocumentBuilder(featureClass, "feature"));
        }
    };
    parallelIndexer.run();
    IndexReader reader = DirectoryReader.open(new RAMDirectory(FSDirectory.open(new File(indexPath)), IOContext.READONCE));
    Bits liveDocs = MultiFields.getLiveDocs(reader);
    double queryCount = 0d;
    ImageSearcher searcher = new GenericFastImageSearcher(100, featureClass, "feature");
    long ms = System.currentTimeMillis();
    for (int i = 0; i < reader.maxDoc(); i++) {
        if (reader.hasDeletions() && !liveDocs.get(i)) continue; // if it is deleted, just ignore it.
        String fileName = getIDfromFileName(reader.document(i).getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]);
        if (queries.keySet().contains(fileName)) {
            queryCount += 1d;
            // ok, we've got a query here for a document ...
            Document queryDoc = reader.document(i);
            ImageSearchHits hits = searcher.search(queryDoc, reader);
        }
    }
    ms = System.currentTimeMillis() - ms;
    System.out.printf("%s \t %3.1f \n", featureClass.getName().substring(featureClass.getName().lastIndexOf('.')+1), (double) ms / queryCount);
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public void testSearchBenchmark() throws IOException {

        System.out.printf("%s\t%s\t%s\t%s\t%s\t%s\n", "Test", "[email protected]", "[email protected]", "[email protected]", "[email protected]", "[email protected]");
        IndexReader reader = DirectoryReader.open(MMapDirectory.open(new File(indexPath)));
        getRecall("ColorLayout (linear)", new GenericFastImageSearcher(50, ColorLayout.class, DocumentBuilder.FIELD_NAME_COLORLAYOUT, true, reader), reader);
        getRecall("PHOG (linear)", new GenericFastImageSearcher(50, PHOG.class, DocumentBuilder.FIELD_NAME_PHOG, true, reader), reader);
        getRecall("JCD (linear)", new GenericFastImageSearcher(50, JCD.class, DocumentBuilder.FIELD_NAME_JCD, true, reader), reader);
        getRecall("EdgeHistogram (linear)", new GenericFastImageSearcher(50, EdgeHistogram.class, DocumentBuilder.FIELD_NAME_EDGEHISTOGRAM, true, reader), reader);
//        getRecall("CEDD (linear)", new GenericFastImageSearcher(50, CEDD.class, DocumentBuilder.FIELD_NAME_CEDD, true, reader), reader);
//        getRecall("JointHistogram (linear)", new GenericFastImageSearcher(50, JointHistogram.class, DocumentBuilder.FIELD_NAME_JOINT_HISTOGRAM, true, reader), reader);
//        getRecall("LocalBinaryPatterns (linear)", new GenericFastImageSearcher(50, LocalBinaryPatterns.class, DocumentBuilder.FIELD_NAME_LOCAL_BINARY_PATTERNS, true, reader), reader);
//        getRecall("Luminance Layout (linear)", new GenericFastImageSearcher(50, LuminanceLayout.class, DocumentBuilder.FIELD_NAME_LUMINANCE_LAYOUT, true, reader), reader);
//        getRecall("BinaryPatternsPyramid (linear)", new GenericFastImageSearcher(50, BinaryPatternsPyramid.class, DocumentBuilder.FIELD_NAME_BINARY_PATTERNS_PYRAMID, true, reader), reader);
//
        getRecall("ColorLayout (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_COLORLAYOUT, DocumentBuilder.FIELD_NAME_COLORLAYOUT + "_hash", new ColorLayout(), 1000), reader);
        getRecall("PHOG (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_PHOG, DocumentBuilder.FIELD_NAME_PHOG + "_hash", new PHOG(), 1000), reader);
        getRecall("JCD (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_JCD, DocumentBuilder.FIELD_NAME_JCD + "_hash", new JCD(), 1000), reader);
        getRecall("EdgeHistogram (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_EDGEHISTOGRAM, DocumentBuilder.FIELD_NAME_EDGEHISTOGRAM + "_hash", new EdgeHistogram(), 1000), reader);
//        getRecall("CEDD (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_CEDD, DocumentBuilder.FIELD_NAME_CEDD + "_hash", new JCD(), 1000), reader);
//        getRecall("JointHistogram (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_JOINT_HISTOGRAM, DocumentBuilder.FIELD_NAME_JOINT_HISTOGRAM + "_hash", new JointHistogram(), 1000), reader);
//        getRecall("LocalBinaryPatterns (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_LOCAL_BINARY_PATTERNS, DocumentBuilder.FIELD_NAME_LOCAL_BINARY_PATTERNS + "_hash", new LocalBinaryPatterns(), 1000), reader);
//        getRecall("Luminance Layout (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_LUMINANCE_LAYOUT, DocumentBuilder.FIELD_NAME_LUMINANCE_LAYOUT + "_hash", new LuminanceLayout(), 1000), reader);
//        getRecall("BinaryPatternsPyramid (hashed)", new BitSamplingImageSearcher(50, DocumentBuilder.FIELD_NAME_BINARY_PATTERNS_PYRAMID, DocumentBuilder.FIELD_NAME_BINARY_PATTERNS_PYRAMID + "_hash", new LocalBinaryPatterns(), 1000), reader);
//        getRecall("VLAD (linear)", new GenericFastImageSearcher(1000, GenericByteLireFeature.class, DocumentBuilder.FIELD_NAME_SURF_VLAD, true, reader), reader);
    }
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public void testColorLayoutFastMap() {
    // creating the list of user objects ...
    LinkedList<LireFeature> objs = new LinkedList<LireFeature>();
    for (Iterator<Document> documentIterator = docs.iterator(); documentIterator.hasNext(); ) {
        Document document = documentIterator.next();
        String[] cls = document.getValues(DocumentBuilder.FIELD_NAME_COLORLAYOUT);
        if (cls.length > 0) {
            ColorLayout clt = new ColorLayout();
            clt.setStringRepresentation(cls[0]);
            objs.add(clt);
        }
    }
    System.out.println("--------------- < COLORLAYOUT > ---------------");
    long nano = System.nanoTime();
    createFastMapForObjects(objs, null);
    nano = System.nanoTime() - nano;
    System.out.println("Time taken: ~ " + (nano / (1000 * 1000 * 1000)) + " s");
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public void testScalableColorFastMap() {
    // creating the list of user objects ...
    LinkedList<LireFeature> objs = new LinkedList<LireFeature>();
    for (Iterator<Document> documentIterator = docs.iterator(); documentIterator.hasNext(); ) {
        Document document = documentIterator.next();
        String[] cls = document.getValues(DocumentBuilder.FIELD_NAME_SCALABLECOLOR);
        if (cls.length > 0) {
            ScalableColor sct = new ScalableColor();
            sct.setStringRepresentation(cls[0]);
            objs.add(sct);
        }
    }
    System.out.println("--------------- < ScalableColor > ---------------");
    long nano = System.nanoTime();
    System.out.println("---------< No cache >----------------");
    createFastMapForObjects(objs, null);
    System.out.println("---------< Array cache >----------------");
    createArrayFastMapForObjects(objs, null);
    nano = System.nanoTime() - nano;
    System.out.println("Time taken: ~ " + (nano / (1000 * 1000)) + " ms");
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public void testEdgeHistogramFastMap() {
    // creating the list of user objects ...
    LinkedList<LireFeature> objs = new LinkedList<LireFeature>();
    for (Iterator<Document> documentIterator = docs.iterator(); documentIterator.hasNext(); ) {
        Document document = documentIterator.next();
        String[] cls = document.getValues(DocumentBuilder.FIELD_NAME_EDGEHISTOGRAM);
        if (cls.length > 0) {
            EdgeHistogram eht = new EdgeHistogram();
            eht.setStringRepresentation(cls[0]);

            objs.add(eht);
        }
    }
    System.out.println("--------------- < EdgeHistogram > ---------------");
    long nano = System.nanoTime();
    createFastMapForObjects(objs, null);
    nano = System.nanoTime() - nano;
    System.out.println("Time taken: ~ " + (nano / (1000 * 1000)) + " ms");
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public void testAutoColorCorrelogramFastMap() {
    // creating the list of user objects ...
    LinkedList<LireFeature> objs = new LinkedList<LireFeature>();
    for (Iterator<Document> documentIterator = docs.iterator(); documentIterator.hasNext(); ) {
        Document document = documentIterator.next();
        String[] cls = document.getValues(DocumentBuilder.FIELD_NAME_AUTOCOLORCORRELOGRAM);
        if (cls.length > 0) {
            AutoColorCorrelogram acc = new AutoColorCorrelogram();
            acc.setStringRepresentation(cls[0]);
            objs.add(acc);
        }
    }
    System.out.println("--------------- < AutoColorCorrelogram > ---------------");
    long nano = System.nanoTime();
    createFastMapForObjects(objs, null);
    nano = System.nanoTime() - nano;
    System.out.println("Time taken: ~ " + (nano / (1000 * 1000)) + " ms");
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
private TopDocs rerank(TopDocs docs, LireFeature feature, IndexReader reader) throws IOException, IllegalAccessException, InstantiationException {
    LireFeature tmp = feature.getClass().newInstance();
    ArrayList<ScoreDoc> res = new ArrayList<ScoreDoc>(docs.scoreDocs.length);
    float maxScore = 0f;
    for (int i = 0; i < docs.scoreDocs.length; i++) {
        tmp.setByteArrayRepresentation(reader.document(docs.scoreDocs[i].doc).getBinaryValue(DocumentBuilder.FIELD_NAME_OPPONENT_HISTOGRAM).bytes);
        maxScore = Math.max(1 / tmp.getDistance(feature), maxScore);
        res.add(new ScoreDoc(docs.scoreDocs[i].doc, 1 / tmp.getDistance(feature)));
    }
    // sorting res ...
    Collections.sort(res, new Comparator<ScoreDoc>() {
        @Override
        public int compare(ScoreDoc o1, ScoreDoc o2) {
            return (int) Math.signum(o2.score - o1.score);
        }
    });
    return new TopDocs(50, (ScoreDoc[]) res.toArray(new ScoreDoc[res.size()]), maxScore);
}
 

import net.semanticmetadata.lire.DocumentBuilder; //导入依赖的package包/类
public void testIndexMissingFiles() throws IOException {
        // first delete some of the existing ones ...
        System.out.println("Deleting visual words from docs ...");
        IndexReader ir = DirectoryReader.open(FSDirectory.open(new File(index)));
        IndexWriter iw = LuceneUtils.createIndexWriter(index, false);
        int maxDocs = ir.maxDoc();
        for (int i = 0; i < maxDocs / 10; i++) {
            Document d = ir.document(i);
            d.removeFields(DocumentBuilder.FIELD_NAME_SURF_VISUAL_WORDS);
            d.removeFields(DocumentBuilder.FIELD_NAME_SURF_LOCAL_FEATURE_HISTOGRAM);
//            d.removeFields(DocumentBuilder.FIELD_NAME_SURF);
            iw.updateDocument(new Term(DocumentBuilder.FIELD_NAME_IDENTIFIER, d.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]), d);
        }
        System.out.println("# of deleted docs:  " + maxDocs / 10);
        System.out.println("Optimizing and closing ...");
        iw.close();
        ir.close();
        System.out.println("Creating new visual words ...");
        SurfFeatureHistogramBuilder surfFeatureHistogramBuilder = new SurfFeatureHistogramBuilder(DirectoryReader.open(FSDirectory.open(new File(index))), numSamples, clusters);
        surfFeatureHistogramBuilder.indexMissing();
        System.out.println("Finished.");
    }