Update order_text_blocks.py

order_text_blocks.py

@@ -84,16 +84,20 @@ def get_hierarchy(labels):
 def textblock_ordering(textblocks, img):
     coords = []
     ori_coords = {}
+    avg_width = 0
     for obj in textblocks:
         x1, y1, x2, y2 = map(float, obj)
         ori_coords[tuple([x1, y1, x2, y2])] = [x1, y1, x2, y2]
         coords.append([x1, y1, x2, y2])
+        avg_width += x2 - x1
+        
+    avg_width /= len(coords)
     
-    # sort the textblocks by y1 
-    coords.sort(key = lambda x: x[1])
-    # create vertical buckets of horizontal pixelsize of 15% of the image width
+    # sort the textblocks by left x coordinate
+    coords.sort(key = lambda x: x[0])
+    # create vertical buckets based on avg. width of text blocks
     buckets = []
-    bucket_size = int(0.15 * img.shape[1])
+    bucket_size = int(0.55 * avg_width)
     # put the textblocks in the buckets
     for coord in coords:
         if len(buckets) == 0:
@@ -106,9 +110,9 @@ def textblock_ordering(textblocks, img):
             else:
                 buckets.append([coord])
                 
-    # sort each bucket by x1
+    # sort each bucket by y1
     for bucket in buckets:
-        bucket.sort(key = lambda x: x[0])
+        bucket.sort(key = lambda x: x[1])
                  
     # visualize the buckets one by one each with a different color
     # for bucket in buckets: