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

Example 6 with SourceImage

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

the class SegmentationTest method testMotlPessiDemKhazns.

/**
 * A very simple, basic page.
 *
 * @throws Exception
 */
@Test
public void testMotlPessiDemKhazns(@Mocked JochrePage jochrePage) throws Exception {
    Map<String, Object> configMap = new HashMap<>();
    configMap.put("jochre.locale", "yi");
    Config config = ConfigFactory.parseMap(configMap).withFallback(ConfigFactory.load());
    JochreSession jochreSession = new JochreSession(config);
    String imageName = "SholemAleykhem_MotelPeysiDemKhazns_12_0.png";
    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(544, 824, 2432, 344);
    Rectangle textPar2 = new Rectangle(552, 1176, 2512, 2112);
    Rectangle textPar3 = new Rectangle(584, 3320, 2448, 344);
    Rectangle textPar4 = new Rectangle(568, 3688, 2464, 592);
    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);
        LOG.debug("realRatio: " + realRatio);
        LOG.debug("expectedRatio: " + expectedRatio);
        if (realRatio > 0.95 && expectedRatio > 0.8) {
            LOG.debug("Found");
            textParagraphs.add(par);
            i++;
        }
        j++;
    }
    assertEquals(textPars.size(), textParagraphs.size());
    int[] rowCounts = new int[] { 3, 18, 3, 5 };
    int[] wordCountsFirstRow = new int[] { 6, 8, 8, 8 };
    for (i = 0; i < textParagraphs.size(); i++) {
        assertEquals(rowCounts[i], textParagraphs.get(i).getRows().size());
        RowOfShapes row = textParagraphs.get(i).getRows().get(0);
        assertEquals(wordCountsFirstRow[i], row.getGroups().size());
    }
}

Example 7 with SourceImage

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

the class SegmentationTest method testAlsacien2.

/**
 * Segmentation errors reported for Alsacien.
 *
 * @throws Exception
 */
@Test
public void testAlsacien2(@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 = "Alsacien2.jpeg";
    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(63, 81, 1059, 108);
    Rectangle textPar2 = new Rectangle(66, 204, 1065, 294);
    Rectangle textPar3 = new Rectangle(63, 516, 1068, 348);
    Rectangle textPar4 = new Rectangle(63, 879, 1071, 537);
    Rectangle textPar5 = new Rectangle(63, 1428, 1068, 354);
    // title paragraph
    textPars.add(textPar1);
    textPars.add(textPar2);
    textPars.add(textPar3);
    textPars.add(textPar4);
    textPars.add(textPar5);
    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[] { 2, 5, 6, 9, 6 };
    int[] wordCountsFirstRow = new int[] { 10, 8, 9, 8, 8 };
    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());
    }
}

Example 8 with SourceImage

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

the class SegmentationTest method testPerets_KhsidisheMayselekh.

/**
 * This page is challenging because of the large blotch of dirt in the lower
 * right. Also, there are several short, indented, one-line paragraphs.
 *
 * @throws Exception
 */
@Test
public void testPerets_KhsidisheMayselekh(@Mocked JochrePage jochrePage) throws Exception {
    Map<String, Object> configMap = new HashMap<>();
    configMap.put("jochre.locale", "yi");
    Config config = ConfigFactory.parseMap(configMap).withFallback(ConfigFactory.load());
    JochreSession jochreSession = new JochreSession(config);
    String imageName = "Peretz_KhsidisheMayselekh_5_0.png";
    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(1670, 920, 624, 182);
    Rectangle textPar2 = new Rectangle(740, 1524, 2368, 96);
    Rectangle textPar3 = new Rectangle(680, 1652, 2620, 192);
    Rectangle textPar4 = new Rectangle(652, 1872, 2648, 800);
    Rectangle textPar5 = new Rectangle(1564, 2696, 1524, 96);
    Rectangle textPar6 = new Rectangle(660, 2820, 2632, 328);
    Rectangle textPar7 = new Rectangle(660, 3176, 2628, 212);
    Rectangle textPar8 = new Rectangle(664, 3404, 2636, 428);
    Rectangle textPar9 = new Rectangle(1992, 3868, 1088, 100);
    Rectangle textPar10 = new Rectangle(664, 4000, 2624, 436);
    Rectangle textPar11 = new Rectangle(664, 4468, 2628, 204);
    // title paragraph
    textPars.add(textPar1);
    textPars.add(textPar2);
    textPars.add(textPar3);
    textPars.add(textPar4);
    textPars.add(textPar5);
    textPars.add(textPar6);
    textPars.add(textPar7);
    textPars.add(textPar8);
    // short paragraph
    textPars.add(textPar9);
    // paragraph with blotch
    textPars.add(textPar10);
    textPars.add(textPar11);
    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);
        LOG.debug("realRatio: " + realRatio);
        LOG.debug("expectedRatio: " + expectedRatio);
        if (realRatio > 0.8 && expectedRatio > 0.8) {
            LOG.debug("Found");
            textParagraphs.add(par);
            i++;
        }
        j++;
    }
    assertEquals(textPars.size(), textParagraphs.size());
    int[] rowCounts = new int[] { 1, 1, 2, 7, 1, 3, 2, 4, 1, 4, 2 };
    int[] wordCountsFirstRow = new int[] { 1, 0, 12, 0, 0, 0, 0, 0, 0, 0, 0 };
    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());
    }
}

Example 9 with SourceImage

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

the class SplitCandidateFinderImplTest method testFindSplitCanidates.

@Test
public void testFindSplitCanidates() throws Exception {
    System.setProperty("config.file", "src/test/resources/test.conf");
    ConfigFactory.invalidateCaches();
    Config config = ConfigFactory.load();
    JochreSession jochreSession = new JochreSession(config);
    InputStream imageFileStream = getClass().getResourceAsStream("shape_370454.png");
    assertNotNull(imageFileStream);
    BufferedImage image = ImageIO.read(imageFileStream);
    final JochrePage page = mock(JochrePage.class);
    JochreImage jochreImage = new SourceImage(page, "name", image, jochreSession);
    Shape shape = jochreImage.getShape(0, 0, jochreImage.getWidth() - 1, jochreImage.getHeight() - 1);
    SplitCandidateFinder splitCandidateFinder = new SplitCandidateFinder(jochreSession);
    List<Split> splits = splitCandidateFinder.findSplitCandidates(shape);
    int[] trueSplitPositions = new int[] { 38, 59, 82 };
    boolean[] foundSplit = new boolean[] { false, false, false };
    for (Split splitCandidate : splits) {
        LOG.debug("Split candidate at " + splitCandidate.getPosition());
        for (int i = 0; i < trueSplitPositions.length; i++) {
            int truePos = trueSplitPositions[i];
            int distance = splitCandidate.getPosition() - truePos;
            if (distance < 0)
                distance = 0 - distance;
            if (distance < splitCandidateFinder.getMinDistanceBetweenSplits()) {
                foundSplit[i] = true;
                LOG.debug("Found split: " + truePos + ", distance " + distance);
            }
        }
    }
    for (int i = 0; i < trueSplitPositions.length; i++) {
        assertTrue("didn't find split " + trueSplitPositions[i], foundSplit[i]);
    }
}

Example 10 with SourceImage

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

the class SegmentationTest method testAlsacien1.

/**
 * Segmentation errors reported for Alsacien.
 *
 * @throws Exception
 */
@Test
public void testAlsacien1(@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 = "Alsacien1.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(715, 517, 462, 115);
    // TODO: for now it's splitting this paragraph by row, since it's assuming
    // paragraphs cannot be
    // both outdented and indented on the same page
    // Rectangle textPar2 = new Rectangle(50, 666, 1798, 1039);
    Rectangle textPar3 = new Rectangle(55, 1837, 1777, 335);
    Rectangle textPar4 = new Rectangle(50, 2211, 1765, 154);
    Rectangle textPar5 = new Rectangle(44, 2404, 1782, 511);
    Rectangle textPar6 = new Rectangle(50, 2948, 1776, 154);
    Rectangle textPar7 = new Rectangle(50, 3135, 1770, 77);
    // title paragraph
    textPars.add(textPar1);
    // textPars.add(textPar2);
    textPars.add(textPar3);
    textPars.add(textPar4);
    textPars.add(textPar5);
    textPars.add(textPar6);
    textPars.add(textPar7);
    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++;
        }
        j++;
    }
    assertEquals(textPars.size(), textParagraphs.size());
    int[] rowCounts = new int[] { 1, 4, 2, 6, 2, 1 };
    int[] wordCountsFirstRow = new int[] { 2, 0, 0, 0, 0, 0, 0 };
    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());
    }
}