more diagnostics for the line finding part of the OCR (pre)process.

GerHobbelt · Jul 27, 2024 · 5608dba · 5608dba
1 parent 5b8c08f
commit 5608dba
Show file tree

Hide file tree

Showing 2 changed files with 52 additions and 4 deletions.
diff --git a/src/textord/linefind.cpp b/src/textord/linefind.cpp
@@ -330,7 +330,7 @@ static void GetLineBoxes(bool horizontal_lines, Image pix_lines, Image pix_inter
 // The output list of TabVector makes no reference to the input BLOBNBOXes.
 void LineFinder::FindLineVectors(const ICOORD &bleft, const ICOORD &tright,
                             BLOBNBOX_LIST *line_bblobs, int *vertical_x, int *vertical_y,
-                            TabVector_LIST *vectors) {
+                            TabVector_LIST *vectors, bool vertical_search) {
   BLOBNBOX_IT bbox_it(line_bblobs);
   int b_count = 0;
   // Put all the blobs into the grid to find the lines, and move the blobs
@@ -364,6 +364,8 @@ void LineFinder::FindLineVectors(const ICOORD &bleft, const ICOORD &tright,
         tprintDebug("Finding line vector starting at bbox ({},{})\n", box.left(), box.bottom());
       }
       AlignedBlobParams align_params(*vertical_x, *vertical_y, box.width());
+      if (!vertical_search)
+        align_params.max_v_gap *= 3;
       TabVector *vector =
           blob_grid.FindVerticalAlignment(align_params, bbox, vertical_x, vertical_y);
       if (vector != nullptr) {
@@ -619,7 +621,7 @@ void LineFinder::FindAndRemoveVLines(Image pix_intersections, int *vertical_x,
   int height = pixGetHeight(src_pix);
   ICOORD bleft(0, 0);
   ICOORD tright(width, height);
-  FindLineVectors(bleft, tright, &line_bblobs, vertical_x, vertical_y, vectors);
+  FindLineVectors(bleft, tright, &line_bblobs, vertical_x, vertical_y, vectors, true);
   if (!vectors->empty()) {
     RemoveUnusedLineSegments(false, &line_bblobs, *pix_vline);
     SubtractLinesAndResidue(*pix_vline, pix_non_vline, src_pix);
@@ -654,7 +656,7 @@ void LineFinder::FindAndRemoveHLines(Image pix_intersections, int vertical_x,
   int height = pixGetHeight(src_pix);
   ICOORD bleft(0, 0);
   ICOORD tright(height, width);
-  FindLineVectors(bleft, tright, &line_bblobs, &vertical_x, &vertical_y, vectors);
+  FindLineVectors(bleft, tright, &line_bblobs, &vertical_x, &vertical_y, vectors, false);
   if (!vectors->empty()) {
     RemoveUnusedLineSegments(true, &line_bblobs, *pix_hline);
     SubtractLinesAndResidue(*pix_hline, pix_non_hline, src_pix);
@@ -701,6 +703,34 @@ void LineFinder::FindAndRemoveLines(int resolution, Image pix, int *vertical_x,
 
   GetLineMasks(resolution, pix, &pix_vline, &pix_non_vline, &pix_hline, &pix_non_hline,
                &pix_intersections, pix_music_mask);
+
+  if (tesseract_->debug_line_finding) {
+#if !GRAPHICS_DISABLED
+    if (!tesseract_->interactive_display_mode && (pix_vline != nullptr || pix_hline != nullptr)) {
+      int width = pixGetWidth(pix);
+      int height = pixGetHeight(pix);
+      ScrollViewReference win = ScrollViewManager::MakeScrollView(tesseract_, "LinesMask", 0, 0, width, height, width, height);
+
+      if (pix_vline != nullptr) {
+        win->Draw(pix_vline, 0, 0, "find & remove H/V lines : vline mask");
+      }
+
+      if (pix_hline != nullptr) {
+        win->Draw(pix_hline, 0, 0, "find & remove H/V lines : hline mask");
+      }
+
+      win->Update();
+    }
+#endif
+
+    if (pix_vline != nullptr) {
+      tesseract_->AddPixDebugPage(pix_vline, "find & remove H/V lines : vline mask");
+    }
+    if (pix_hline != nullptr) {
+      tesseract_->AddPixDebugPage(pix_hline, "find & remove H/V lines : hline mask");
+    }
+  }
+
   // Find lines, convert to TabVector_LIST and remove those that are used.
   FindAndRemoveVLines(pix_intersections, vertical_x, vertical_y, &pix_vline,
                       pix_non_vline, pix, v_lines);
@@ -718,6 +748,24 @@ void LineFinder::FindAndRemoveLines(int resolution, Image pix, int *vertical_x,
   FindAndRemoveHLines(pix_intersections, *vertical_x, *vertical_y, &pix_hline,
                       pix_non_hline, pix, h_lines);
   if (tesseract_->debug_line_finding) {
+#if !GRAPHICS_DISABLED
+    if (!tesseract_->interactive_display_mode && (pix_vline != nullptr || pix_hline != nullptr)) {
+      int width = pixGetWidth(pix);
+      int height = pixGetHeight(pix);
+      ScrollViewReference win = ScrollViewManager::MakeScrollView(tesseract_, "LinesMask", 0, 0, width, height, width, height);
+
+      if (pix_vline != nullptr) {
+        win->Draw(pix_vline, 0, 0, "find & remove H/V lines : vline");
+      }
+
+      if (pix_hline != nullptr) {
+        win->Draw(pix_hline, 0, 0, "find & remove H/V lines : hline");
+      }
+
+      win->Update();
+    }
+#endif
+
     if (pix_vline != nullptr) {
       tesseract_->AddPixDebugPage(pix_vline, "find & remove H/V lines : vline");
     }

diff --git a/src/textord/linefind.h b/src/textord/linefind.h
@@ -85,7 +85,7 @@ class LineFinder {
   // The output list of TabVector makes no reference to the input BLOBNBOXes.
   void FindLineVectors(const ICOORD& bleft, const ICOORD& tright,
                               BLOBNBOX_LIST* line_bblobs, int* vertical_x, int* vertical_y,
-                              TabVector_LIST* vectors);
+                              TabVector_LIST* vectors, bool vertical_search);
 
   // Finds vertical line objects in pix_vline and removes them from src_pix.
   // Uses the given resolution to determine size thresholds instead of any