public byte[] getByteArrayRepresentation() {
byte[] result = new byte[correlogram.length * correlogram[0].length * 4 + 5];
for (int i = 0; i < correlogram.length; i++) {
System.arraycopy(
SerializationUtils.toByteArray(correlogram[i]),
0,
result,
i * correlogram[i].length * 4,
correlogram[i].length * 4);
}
System.arraycopy(SerializationUtils.toBytes(numBins), 0, result, result.length - 5, 4);
result[result.length - 1] = (byte) distanceSet.length;
return result;
}
public void setByteArrayRepresentation(byte[] in) {
byte[] numBinsBytes = new byte[4];
numBinsBytes[0] = in[in.length - 5];
numBinsBytes[1] = in[in.length - 4];
numBinsBytes[2] = in[in.length - 3];
numBinsBytes[3] = in[in.length - 2];
int maxDistance = (int) in[in.length - 1];
numBins = SerializationUtils.toInt(numBinsBytes);
correlogram = new float[numBins][maxDistance];
float[] temp = SerializationUtils.toFloatArray(in);
for (int i = 0; i < correlogram.length; i++) {
System.arraycopy(temp, i * maxDistance, correlogram[i], 0, maxDistance);
}
distanceSet = new int[maxDistance];
for (int i = 0; i < distanceSet.length; i++) {
distanceSet[i] = i + 1;
}
}
/**
* Reads data from a file and writes it to an index.
*
* @param indexWriter the index to write to.
* @param inputFile the input data for the process.
* @throws IOException
* @throws InstantiationException
* @throws IllegalAccessException
* @throws ClassNotFoundException
*/
private void readFile(IndexWriter indexWriter, File inputFile)
throws IOException, InstantiationException, IllegalAccessException, ClassNotFoundException {
BufferedInputStream in = new BufferedInputStream(new FileInputStream(inputFile));
byte[] tempInt = new byte[4];
int tmp, tmpFeature, count = 0;
byte[] temp = new byte[100 * 1024];
// read file hashFunctionsFileName length:
while (in.read(tempInt, 0, 4) > 0) {
Document d = new Document();
tmp = SerializationUtils.toInt(tempInt);
// read file hashFunctionsFileName:
in.read(temp, 0, tmp);
String filename = new String(temp, 0, tmp);
// normalize Filename to full path.
filename =
inputFile
.getCanonicalPath()
.substring(0, inputFile.getCanonicalPath().lastIndexOf(inputFile.getName()))
+ filename;
d.add(new StringField(DocumentBuilder.FIELD_NAME_IDENTIFIER, filename, Field.Store.YES));
// System.out.print(filename);
while ((tmpFeature = in.read()) < 255) {
// System.out.print(", " + tmpFeature);
LireFeature f = (LireFeature) Class.forName(Extractor.features[tmpFeature]).newInstance();
// byte[] length ...
in.read(tempInt, 0, 4);
tmp = SerializationUtils.toInt(tempInt);
// read feature byte[]
in.read(temp, 0, tmp);
f.setByteArrayRepresentation(temp, 0, tmp);
addToDocument(f, d, Extractor.featureFieldNames[tmpFeature]);
// d.add(new StoredField(Extractor.featureFieldNames[tmpFeature],
// f.getByteArrayRepresentation()));
}
if (run == 2) indexWriter.addDocument(d);
docCount++;
// if (count%1000==0) System.out.print('.');
// if (count%10000==0) System.out.println(" " + count);
}
in.close();
}
protected void addToDocument(GlobalFeature feature, Document document, String featureFieldName) {
// This is for debugging the image features.
// try {
//// System.out.println(feature.getClass().getName() + " " +
// document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0]);
// LireFeature f1 = feature.getClass().newInstance();
// f1.extract(ImageIO.read(new
// File(document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0])));
// float distance = feature.getDistance(f1);
// if (distance != 0) {
// System.out.println("Extracted:" +
// java.util.Arrays.toString(f1.getFeatureVector()).replaceAll("\\.0,", "") + "\n" +
// "Data :" +
// java.util.Arrays.toString(feature.getFeatureVector()).replaceAll("\\.0,", "") + "\n" +
// "Problem with " + f1.getClass().getName() + " at file " +
// document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0] + ", distance=" + distance
// );
//// System.out.println("Problem with " + f1.getClass().getName() + " at file " +
// document.getValues(DocumentBuilder.FIELD_NAME_IDENTIFIER)[0] + ", distance=" + distance);
// }
// } catch (Exception e) {
// e.printStackTrace();
//
// }
if (feature.getClass().getCanonicalName().equals(featureClass.getCanonicalName())) {
// generate hashes here:
// int[] hashes =
// LocalitySensitiveHashing.generateHashes(feature.getFeatureVector());
int[] hashes = BitSampling.generateHashes(feature.getFeatureVector());
// System.out.println(Arrays.toString(hashes));
// store hashes in index as terms
document.add(
new TextField(
featureFieldName + "_hash",
SerializationUtils.arrayToString(hashes),
Field.Store.YES));
// add the specific feature
document.add(new StoredField(featureFieldName, feature.getByteArrayRepresentation()));
}
// add the specific feature
// document.add(new StoredField(featureFieldName, feature.getByteArrayRepresentation()));
}
@Override
public void setByteArrayRepresentation(byte[] in, int offset, int length) {
histogram = SerializationUtils.toDoubleArray(in, offset, length);
}
public void setByteArrayRepresentation(byte[] in) {
histogram = SerializationUtils.toDoubleArray(in);
}
public byte[] getByteArrayRepresentation() {
return SerializationUtils.toByteArray(histogram);
}