bounding box polishing

kermitt2 · Oct 28, 2015 · 2098204 · 2098204
1 parent 3f4201c
commit 2098204
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 4 deletions.
diff --git a/grobid-core/src/main/java/org/grobid/core/layout/BoundingBox.java b/grobid-core/src/main/java/org/grobid/core/layout/BoundingBox.java
@@ -21,7 +21,7 @@ private BoundingBox(int page, double x, double y, double width, double height) {
     }
 
     public static BoundingBox fromTwoPoints(int page, double x1, double y1, double x2, double y2) {
-        if (x1 >= x2 || y1 > y2) {
+        if (x1 > x2 || y1 > y2) {
             throw new IllegalArgumentException("Invalid points provided: (" + x1 + ";" + y1 + ")-(" + x2 + ";" + y2 + ")");
         }
         return new BoundingBox(page, x1, y1, x2 - x1, y2 - y1);

diff --git a/grobid-core/src/main/java/org/grobid/core/utilities/BoundingBoxCalculator.java b/grobid-core/src/main/java/org/grobid/core/utilities/BoundingBoxCalculator.java
@@ -11,8 +11,8 @@
  * Utilities to calculate bounding boxes from coordinates
  */
 public class BoundingBoxCalculator {
-    private static final double EPS_X = 10;
-    private static final double EPS_Y = 3;
+    private static final double EPS_X = 15;
+    private static final double EPS_Y = 4;
 
     public static List<BoundingBox> calculate(List<LayoutToken> tokens) {
         List<BoundingBox> result = Lists.newArrayList();
@@ -26,6 +26,9 @@ public static List<BoundingBox> calculate(List<LayoutToken> tokens) {
         BoundingBox lastBox = firstBox;
         for (int i = 1; i < tokens.size(); i++) {
             BoundingBox b = BoundingBox.fromLayoutToken(tokens.get(i));
+            if (Math.abs(b.getWidth()) <= Double.MIN_VALUE || Math.abs(b.getHeight()) <= Double.MIN_VALUE) {
+                continue;
+            }
 
             if (near(lastBox, b)) {
                 result.set(result.size() - 1, result.get(result.size() - 1).boundBox(b));
@@ -37,8 +40,11 @@ public static List<BoundingBox> calculate(List<LayoutToken> tokens) {
         return result;
     }
 
+    //same page, Y is more or less the same, b2 follows b1 on X, and b2 close to the end of b1
     private static boolean near(BoundingBox b1, BoundingBox b2) {
-        return Math.abs(b1.getY() - b2.getY()) < EPS_Y && Math.abs(b1.getY2() - b2.getY2()) < EPS_Y && b2.getX() - b1.getX2() < EPS_X;
+        return b1.getPage() == b2.getPage()
+                && Math.abs(b1.getY() - b2.getY()) < EPS_Y && Math.abs(b1.getY2() - b2.getY2()) < EPS_Y
+                && b2.getX() - b1.getX2() < EPS_X && b2.getX() >= b1.getX();
     }
 
 }