Examples with SourceImage com.joliciel.jochre.graphics.SourceImage used on opensource projects

Example 11 with SourceImage

use of com.joliciel.jochre.graphics.SourceImage in project jochre by urieli.

the class SegmentationTest method testAlsacienPlay3.

/**
 * Segmentation errors reported for Alsacien play - challenging because of the
 * unusual indentation.
 *
 * @throws Exception
 */
@Test
public void testAlsacienPlay3(@Mocked final JochrePage jochrePage, @Mocked final JochreDocument jochreDoc) throws Exception {
    Map<String, Object> configMap = new HashMap<>();
    configMap.put("jochre.locale", "de");
    Config config = ConfigFactory.parseMap(configMap).withFallback(ConfigFactory.load());
    JochreSession jochreSession = new JochreSession(config);
    new Expectations() {

        {
            jochrePage.getDocument();
            result = jochreDoc;
            minTimes = 0;
            jochreDoc.isLeftToRight();
            result = true;
            minTimes = 0;
        }
    };
    String imageName = "AlsacienPlay3.jpg";
    LOG.debug(imageName);
    InputStream imageFileStream = getClass().getResourceAsStream("/com/joliciel/jochre/segmentation/" + imageName);
    assertNotNull(imageFileStream);
    BufferedImage image = ImageIO.read(imageFileStream);
    SourceImage sourceImage = new SourceImage(jochrePage, "", image, jochreSession);
    Segmenter segmenter = new Segmenter(sourceImage, jochreSession);
    segmenter.segment();
    List<Rectangle> textPars = new ArrayList<>();
    Rectangle textPar1 = new Rectangle(712, 532, 556, 52);
    Rectangle textPar2 = new Rectangle(324, 600, 1324, 128);
    Rectangle textPar3 = new Rectangle(680, 730, 592, 50);
    Rectangle textPar4 = new Rectangle(404, 808, 684, 48);
    // title paragraph
    textPars.add(textPar1);
    textPars.add(textPar2);
    textPars.add(textPar3);
    textPars.add(textPar4);
    int i = 0;
    int j = 0;
    List<Paragraph> textParagraphs = new ArrayList<>();
    for (Paragraph par : sourceImage.getParagraphs()) {
        Rectangle real = new Rectangle(par.getLeft(), par.getTop(), par.getRight() - par.getLeft(), par.getBottom() - par.getTop());
        Rectangle expected = textPars.get(i);
        Rectangle intersection = expected.intersection(real);
        double realArea = real.width * real.height;
        double expectedArea = expected.width * expected.height;
        double intersectionArea = intersection.width * intersection.height;
        double realRatio = intersectionArea / realArea;
        double expectedRatio = intersectionArea / expectedArea;
        LOG.debug("Paragraph " + j + ": " + par.toString());
        LOG.debug("realRatio: " + realRatio);
        LOG.debug("expectedRatio: " + expectedRatio);
        if (realRatio > 0.8 && expectedRatio > 0.8) {
            LOG.debug("Found");
            textParagraphs.add(par);
            i++;
            if (i >= textPars.size())
                break;
        }
        j++;
    }
    assertEquals(textPars.size(), textParagraphs.size());
    int[] rowCounts = new int[] { 1, 2, 1, 1 };
    // TODO: words in "spaced" rows (uses spacing to emphasize instead of bold
    // or italics) get split
    // should try to detect multiple single letter words
    int[] wordCountsFirstRow = new int[] { 0, 10, 0, 5 };
    for (i = 0; i < textParagraphs.size(); i++) {
        assertEquals("row count " + i, rowCounts[i], textParagraphs.get(i).getRows().size());
        RowOfShapes row = textParagraphs.get(i).getRows().get(0);
        if (wordCountsFirstRow[i] > 0)
            assertEquals("word count " + i, wordCountsFirstRow[i], row.getGroups().size());
    }
}