Support image width and height larger than 32767

Builds now can define LARGE_IMAGES to use 32 bit for image dimensions instead of 16 bit. Signed-off-by: Stefan Weil <[email protected]>
tesseract-ocr · Oct 27, 2021 · 2e508b3 · 2e508b3
1 parent 59fbad0
commit 2e508b3
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 9 deletions.
diff --git a/src/ccstruct/mod128.cpp b/src/ccstruct/mod128.cpp
@@ -20,7 +20,7 @@
 
 namespace tesseract {
 
-static const int16_t idirtab[] = {
+static const TDimension idirtab[] = {
     1000, 0,    998,  49,   995,  98,   989,  146,  980,   195,  970,  242,  956,  290,   941,
     336,  923,  382,  903,  427,  881,  471,  857,  514,   831,  555,  803,  595,  773,   634,
     740,  671,  707,  707,  671,  740,  634,  773,  595,   803,  555,  831,  514,  857,   471,

diff --git a/src/ccstruct/points.cpp b/src/ccstruct/points.cpp
@@ -40,12 +40,32 @@ bool FCOORD::normalise() { // Convert to unit vec
   return true;
 }
 
+// Deserialize an ICOORD.
+// For compatibility reasons it uses unsigned 16 bit coordinates
+// instead of 32 bit coordinates.
 bool ICOORD::DeSerialize(TFile *f) {
-  return f->DeSerialize(&xcoord) && f->DeSerialize(&ycoord);
+  bool success = false;
+  uint16_t coord;
+  if (f->DeSerialize(&coord)) {
+    xcoord = coord;
+    if (f->DeSerialize(&coord)) {
+      ycoord = coord;
+      success = true;
+    }
+  }
+  return success;
 }
 
+// Serialize an ICOORD.
+// For compatibility reasons it uses unsigned 16 bit coordinates
+// instead of 32 bit coordinates.
 bool ICOORD::Serialize(TFile *f) const {
-  return f->Serialize(&xcoord) && f->Serialize(&ycoord);
+  uint16_t coord;
+  coord = xcoord;
+  auto success = f->Serialize(&coord);
+  coord = ycoord;
+  success = success && f->Serialize(&coord);
+  return success;
 }
 
 // Set from the given x,y, shrinking the vector to fit if needed.

diff --git a/src/textord/tordmain.cpp b/src/textord/tordmain.cpp
@@ -209,12 +209,8 @@ void assign_blobs_to_blocks2(Image pix,
  **********************************************************************/
 
 void Textord::find_components(Image pix, BLOCK_LIST *blocks, TO_BLOCK_LIST *to_blocks) {
-  int width = pixGetWidth(pix);
-  int height = pixGetHeight(pix);
-  if (width > INT16_MAX || height > INT16_MAX) {
-    tprintf("Input image too large! (%d, %d)\n", width, height);
-    return; // Can't handle it.
-  }
+  auto width = pixGetWidth(pix);
+  auto height = pixGetHeight(pix);
 
   BLOCK_IT block_it(blocks); // iterator
   for (block_it.mark_cycle_pt(); !block_it.cycled_list(); block_it.forward()) {