Merge pull request #1329 from myhloli/dev

feat(pre_proc): add function to remove overlapping characters in spans

magic_pdf/pdf_parse_union_core_v2.py

@@ -34,7 +34,8 @@ from magic_pdf.pre_proc.construct_page_dict import ocr_construct_page_component_
 from magic_pdf.pre_proc.cut_image import ocr_cut_image_and_table
 from magic_pdf.pre_proc.ocr_detect_all_bboxes import ocr_prepare_bboxes_for_layout_split_v2
 from magic_pdf.pre_proc.ocr_dict_merge import fill_spans_in_blocks, fix_block_spans_v2, fix_discarded_block
-from magic_pdf.pre_proc.ocr_span_list_modify import get_qa_need_list_v2, remove_overlaps_low_confidence_spans, remove_overlaps_min_spans
+from magic_pdf.pre_proc.ocr_span_list_modify import get_qa_need_list_v2, remove_overlaps_low_confidence_spans, \
+    remove_overlaps_min_spans, remove_overlaps_chars
 
 os.environ['NO_ALBUMENTATIONS_UPDATE'] = '1'  # 禁止albumentations检查更新
 
@@ -120,6 +121,10 @@ def fill_char_in_spans(spans, all_chars):
     empty_spans = []
 
     for span in spans:
+
+        # 移除同一个span中重叠的char
+        span['chars'] = remove_overlaps_chars(span['chars'])
+
         chars_to_content(span)
         # 有的span中虽然没有字但有一两个空的占位符，用宽高和content长度过滤
         if len(span['content']) * span['height'] < span['width'] * 0.5:

magic_pdf/pre_proc/ocr_span_list_modify.py

@@ -33,6 +33,31 @@ def remove_overlaps_low_confidence_spans(spans):
     return spans, dropped_spans
 
 
+def remove_overlaps_chars(chars):
+    dropped_chars = []
+    #  删除重叠的char
+    for char1 in chars:
+        for char2 in chars:
+            if char1 != char2:
+                # char1 或 char2 任何一个都不应该在 dropped_chars 中
+                if char1 in dropped_chars or char2 in dropped_chars:
+                    continue
+                else:
+                    if calculate_iou(char1['bbox'], char2['bbox']) > 0.95:
+                        char_need_remove = char1
+                        if (
+                            char_need_remove is not None
+                            and char_need_remove not in dropped_chars
+                        ):
+                            dropped_chars.append(char_need_remove)
+
+    if len(dropped_chars) > 0:
+        for char_need_remove in dropped_chars:
+            chars.remove(char_need_remove)
+
+    return chars
+
+
 def remove_overlaps_min_spans(spans):
     dropped_spans = []
     #  删除重叠spans中较小的那些