fix: update figure caption match algorithm

magic_pdf/model/magic_model.py

@@ -519,15 +519,27 @@ class MagicModel:
             if nxt is None:
                 break
     
-            seen_idx.add(fst_idx)
-            seen_idx.add(nxt[0])
             if fst_kind == SUB_BIT_KIND:
-                seen_sub_idx.add(fst_idx)
                 sub_idx, obj_idx = fst_idx, nxt[0] - OBJ_IDX_OFFSET
 
             else:
-                seen_sub_idx.add(nxt[0])
                 sub_idx, obj_idx = nxt[0], fst_idx - OBJ_IDX_OFFSET
+                
+            pair_dis = bbox_distance(subjects[sub_idx]['bbox'], objects[obj_idx]['bbox'])
+            nearest_dis = float('inf')
+            for i in range(N):
+                if i in seen_idx:continue
+                nearest_dis = min(nearest_dis, bbox_distance(subjects[i]['bbox'], objects[obj_idx]['bbox']))
+
+            if pair_dis >= 3*nearest_dis:
+                seen_idx.add(sub_idx)
+                continue 
+
+            
+            seen_idx.add(sub_idx)
+            seen_idx.add(obj_idx + OBJ_IDX_OFFSET)
+            seen_sub_idx.add(sub_idx)
+            
             ret.append(
                 {
                     'sub_bbox': {