Marmelator/layout-parser
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 289 Wiki Activity

REmove more

Browse Source
This commit is contained in:
Dominique Eifländer 2024-02-15 10:29:39 +01:00
parent 4afa8daafa
commit 4de6c12aec
7 changed files with 7 additions and 214 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

168
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/docstrum/DocstrumSegmenter.java
View File

@ -14,7 +14,6 @@ import com.google.common.collect.Lists;
import com.knecon.fforesight.service.layoutparser.processor.model.text.TextPositionSequence;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.DisjointSets;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.Histogram;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.BoundingBox;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.Line;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.Word;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.Zone;
@ -27,8 +26,6 @@ import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.mo
public class DocstrumSegmenter {
public static final int MAX_ZONES_PER_PAGE = 300;
public static final double ORIENTATION_MARGIN = 0.2;
public static final int LINES_PER_PAGE_MARGIN = 100;
private static final double DISTANCE_STEP = 16.0;
@ -181,21 +178,7 @@ public class DocstrumSegmenter {
double characterSpacing = computeCharacterSpacing(components, orientation);
double lineSpacing = computeLineSpacing(components, orientation);
List<ComponentLine> lines = determineLines(components, orientation, characterSpacing * COMP_DIST_CHAR, lineSpacing * MAX_VERTICAL_COMP_DIST);
if (Math.abs(orientation) > ORIENTATION_MARGIN) {
List<ComponentLine> linesZero = determineLines(components, 0, characterSpacing * COMP_DIST_CHAR, lineSpacing * MAX_VERTICAL_COMP_DIST);
if (Math.abs(lines.size() - LINES_PER_PAGE_MARGIN) > Math.abs(linesZero.size() - LINES_PER_PAGE_MARGIN)) {
orientation = 0;
lines = linesZero;
}
}
double lineOrientation = computeOrientation(lines);
if (!Double.isNaN(lineOrientation)) {
orientation = lineOrientation;
}
List<ComponentLine> lines = determineLines(components, characterSpacing * COMP_DIST_CHAR, lineSpacing * MAX_VERTICAL_COMP_DIST);
List<List<ComponentLine>> zones = determineZones(lines,
orientation,
@ -207,19 +190,10 @@ public class DocstrumSegmenter {
0.0,
0.0,
lineSpacing * MAX_VERTICAL_MERGE_DIST);
zones = mergeZones(zones, characterSpacing * 0.5);
zones = mergeLines(zones, orientation, Double.NEGATIVE_INFINITY, 0.0, 0.0, lineSpacing * MAX_VERTICAL_MERGE_DIST);
return convertToBxModel(zones, WORD_DIST_MULT * characterSpacing);
}
/**
* Performs for each component search for nearest-neighbors and stores the
* result in component's neighbors attribute.
*
* @param components array of components
* equal to the number of nearest-neighbors per component.
*/
private void findNeighbors(Character[] components) {
if (components.length == 0) {
@ -352,15 +326,14 @@ public class DocstrumSegmenter {
* Groups components into text lines.
*
* @param components component list
* @param orientation - estimated text orientation
* @param maxHorizontalDistance - maximum horizontal distance between components
* @param maxVerticalDistance - maximum vertical distance between components
* @return lines of components
*/
private List<ComponentLine> determineLines(List<Character> components, double orientation, double maxHorizontalDistance, double maxVerticalDistance) {
private List<ComponentLine> determineLines(List<Character> components, double maxHorizontalDistance, double maxVerticalDistance) {
DisjointSets<Character> sets = new DisjointSets<Character>(components);
AngleFilter filter = AngleFilter.newInstance(orientation - ANGLE_TOLERANCE, orientation + ANGLE_TOLERANCE);
AngleFilter filter = AngleFilter.newInstance(-ANGLE_TOLERANCE, ANGLE_TOLERANCE);
for (Character component : components) {
for (Neighbor neighbor : component.getNeighbors()) {
double x = neighbor.getHorizontalDistance() / maxHorizontalDistance;
@ -374,24 +347,12 @@ public class DocstrumSegmenter {
for (Set<Character> group : sets) {
List<Character> lineComponents = new ArrayList<Character>(group);
Collections.sort(lineComponents, Character.CharacterXComparator.getInstance());
lines.add(new ComponentLine(lineComponents, orientation));
lines.add(new ComponentLine(lineComponents));
}
return lines;
}
private double computeOrientation(List<ComponentLine> lines) {
// Compute weighted mean of line angles
double valueSum = 0.0;
double weightSum = 0.0;
for (ComponentLine line : lines) {
valueSum += line.getAngle() * line.getLength();
weightSum += line.getLength();
}
return valueSum / weightSum;
}
private List<List<ComponentLine>> determineZones(List<ComponentLine> lines,
double orientation,
double minHorizontalDistance,
@ -444,111 +405,6 @@ public class DocstrumSegmenter {
}
private List<List<ComponentLine>> mergeZones(List<List<ComponentLine>> zones, double tolerance) {
List<BoundingBox> bounds = new ArrayList<BoundingBox>(zones.size());
for (List<ComponentLine> zone : zones) {
BoundingBoxBuilder builder = new BoundingBoxBuilder();
for (ComponentLine line : zone) {
for (Character component : line.getComponents()) {
builder.expand(component.getChunk());
}
}
bounds.add(builder.getBounds());
}
List<List<ComponentLine>> outputZones = new ArrayList<List<ComponentLine>>();
mainFor:
for (int i = 0; i < zones.size(); i++) {
for (int j = 0; j < zones.size(); j++) {
if (i == j || bounds.get(j) == null || bounds.get(i) == null) {
continue;
}
if (bounds.get(j).contains(bounds.get(i), tolerance)) {
zones.get(j).addAll(zones.get(i));
bounds.set(i, null);
continue mainFor;
}
}
outputZones.add(zones.get(i));
}
return outputZones;
}
private List<List<ComponentLine>> mergeLines(List<List<ComponentLine>> zones,
double orientation,
double minHorizontalDistance,
double maxHorizontalDistance,
double minVerticalDistance,
double maxVerticalDistance) {
List<List<ComponentLine>> outputZones = new ArrayList<List<ComponentLine>>(zones.size());
for (List<ComponentLine> zone : zones) {
outputZones.add(mergeLinesInZone(zone, orientation, minHorizontalDistance, maxHorizontalDistance, minVerticalDistance, maxVerticalDistance));
}
return outputZones;
}
private List<ComponentLine> mergeLinesInZone(List<ComponentLine> lines,
double orientation,
double minHorizontalDistance,
double maxHorizontalDistance,
double minVerticalDistance,
double maxVerticalDistance) {
DisjointSets<ComponentLine> sets = new DisjointSets<ComponentLine>(lines);
for (int i = 0; i < lines.size(); i++) {
ComponentLine li = lines.get(i);
for (int j = i + 1; j < lines.size(); j++) {
ComponentLine lj = lines.get(j);
double hDist = li.horizontalDistance(lj, orientation);
double vDist = li.verticalDistance(lj, orientation);
if (minHorizontalDistance <= hDist && hDist <= maxHorizontalDistance && minVerticalDistance <= vDist && vDist <= maxVerticalDistance) {
sets.union(li, lj);
} else if (minVerticalDistance <= vDist && vDist <= maxVerticalDistance && Math.abs(hDist - Math.min(li.getLength(), lj.getLength())) < 0.1) {
boolean componentOverlap = false;
int overlappingCount = 0;
for (Character ci : li.getComponents()) {
for (Character cj : lj.getComponents()) {
double dist = ci.overlappingDistance(cj, orientation);
if (dist > 2) {
componentOverlap = true;
}
if (dist > 0) {
overlappingCount++;
}
}
}
if (!componentOverlap && overlappingCount <= 2) {
sets.union(li, lj);
}
}
}
}
List<ComponentLine> outputZone = new ArrayList<ComponentLine>();
for (Set<ComponentLine> group : sets) {
List<Character> components = new ArrayList<Character>();
for (ComponentLine line : group) {
components.addAll(line.getComponents());
}
Collections.sort(components, Character.CharacterXComparator.getInstance());
outputZone.add(new ComponentLine(components, orientation));
}
return outputZone;
}
/**
* Converts list of zones from internal format (using components and
* component lines) to BxModel.
*
* @param zones zones in internal format
* @param wordSpacing - maximum allowed distance between components that
* belong to one word
* @return BxModel page
*/
private List<Zone> convertToBxModel(List<List<ComponentLine>> zones, double wordSpacing) {
List<Zone> zoneList = new ArrayList<>();
@ -630,7 +486,7 @@ public class DocstrumSegmenter {
private final List<Character> components;
public ComponentLine(List<Character> components, double orientation) {
public ComponentLine(List<Character> components) {
this.components = components;
@ -653,7 +509,7 @@ public class DocstrumSegmenter {
} else if (!components.isEmpty()) {
Character component = components.get(0);
double dx = component.getChunk().getWidthDirAdj() / 3;
double dy = dx * Math.tan(orientation);
double dy = dx * Math.tan(0);
this.x0 = component.getX() - dx;
this.x1 = component.getX() + dx;
this.y0 = component.getY() - dy;
@ -671,12 +527,6 @@ public class DocstrumSegmenter {
}
public double getSlope() {
return (y1 - y0) / (x1 - x0);
}
public double getLength() {
return Math.sqrt((x0 - x1) * (x0 - x1) + (y0 - y1) * (y0 - y1));
@ -699,12 +549,6 @@ public class DocstrumSegmenter {
}
public List<Character> getComponents() {
return components;
}
public double angularDifference(ComponentLine j) {
double diff = Math.abs(getAngle() - j.getAngle());

29
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/docstrum/HierarchicalReadingOrderResolver.java
View File

@ -8,11 +8,7 @@ import java.util.List;
import org.springframework.stereotype.Service;
import com.google.common.collect.Lists;
import com.knecon.fforesight.service.layoutparser.processor.model.text.RedTextPosition;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.BBoxObject;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.Line;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.Word;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.Zone;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.model.refactor.utils.DoubleUtils;
import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.readingorder.BBoxZoneGroup;
@ -24,7 +20,6 @@ import com.knecon.fforesight.service.layoutparser.processor.services.docstrum.re
public class HierarchicalReadingOrderResolver {
static final int GRIDSIZE = 50;
static final double BOXES_FLOW = 0.5;
static final double EPS = 0.01;
static final int MAX_ZONES = 1000;
static final Comparator<BBoxObject> Y_ASCENDING_ORDER = new Comparator<BBoxObject>() {
@ -45,15 +40,6 @@ public class HierarchicalReadingOrderResolver {
}
};
static final Comparator<RedTextPosition> TP_X_ASCENDING_ORDER = new Comparator<RedTextPosition>() {
@Override
public int compare(RedTextPosition o1, RedTextPosition o2) {
return DoubleUtils.compareDouble(o1.getXDirAdj(), o2.getXDirAdj(), EPS);
}
};
static final Comparator<BBoxObject> YX_ASCENDING_ORDER = new Comparator<BBoxObject>() {
@Override
@ -67,21 +53,6 @@ public class HierarchicalReadingOrderResolver {
public List<Zone> resolve(List<Zone> zones) {
for (Zone zone : zones) {
List<Line> lines = Lists.newArrayList(zone.getLines());
for (Line line : lines) {
List<Word> words = Lists.newArrayList(line.getWords());
for (Word word : words) {
List<RedTextPosition> chunks = Lists.newArrayList(word.getTextPositions());
Collections.sort(chunks, TP_X_ASCENDING_ORDER);
word.setTextPositions(chunks);
}
Collections.sort(words, X_ASCENDING_ORDER);
line.setWords(words);
}
Collections.sort(lines, YX_ASCENDING_ORDER);
zone.setLines(lines);
}
List<Zone> orderedZones;
if (zones.size() > MAX_ZONES) {
orderedZones = new ArrayList<Zone>(zones);

22
layoutparser-service/layoutparser-service-processor/src/main/java/com/knecon/fforesight/service/layoutparser/processor/services/docstrum/readingorder/DocumentPlane.java
View File

@ -255,26 +255,4 @@ public class DocumentPlane {
return objs_.size();
}
public String dump() {
StringBuilder sb = new StringBuilder();
for (GridXY iter : grid.keySet()) {
sb.append(iter.toString()).append(" [");
for (BBoxObject obj : grid.get(iter)) {
if (obj instanceof BBoxZoneGroup) {
BBoxZoneGroup group = (BBoxZoneGroup) obj;
sb.append(group.getLeftChild());
sb.append(group.getRightChild());
} else if (obj instanceof Zone) {
Zone zone = (Zone) obj;
sb.append(zone);
}
sb.append("\n");
}
sb.append("]\n");
}
return sb.toString();
}
}

2
layoutparser-service/layoutparser-service-server/src/test/java/com/knecon/fforesight/service/layoutparser/server/graph/ViewerDocumentTest.java
View File

@ -25,7 +25,7 @@ public class ViewerDocumentTest extends BuildDocumentTest {
@SneakyThrows
public void testViewerDocument() {
String fileName = "files/new/agb1.pdf";
String fileName = "files/Plenarprotokoll 1 (keine Druchsache!) (1).pdf";
String tmpFileName = "/tmp/" + Path.of(fileName).getFileName() + "_VIEWER.pdf";
var documentFile = new ClassPathResource(fileName).getFile();

BIN
layoutparser-service/layoutparser-service-server/src/test/resources/files/Cyberport__SD-Faktura-Kopie_(ZRG2)_-_03.09.2021.pdf Normal file
View File

Binary file not shown.

BIN
layoutparser-service/layoutparser-service-server/src/test/resources/files/brokenTableOnOcr_ocred.pdf Normal file
View File

Binary file not shown.

BIN
layoutparser-service/layoutparser-service-server/src/test/resources/files/cyberport Rechnung.pdf Normal file
View File

Binary file not shown.