qurator-spk
/
sbb_textline_detection
mirror of https://github.com/qurator-spk/sbb_textline_detection.git
1
0
Fork 0
Code Issues Releases Wiki Activity

Update main.py

Browse Source
pull/1/head
Rezanezhad, Vahid 5 years ago
parent 0b7bc8d93e
commit 47d972b459
1 changed files with 100 additions and 95 deletions
Show Stats Download Patch File Download Diff File

159
qurator/sbb_textline_detector/main.py
Unescape Escape View File

@ -1374,103 +1374,102 @@ class textlineerkenner:
coord_page.set('points',points_page_print) coord_page.set('points',points_page_print)
""" """
if len(contours)>0:
region_order=ET.SubElement(page, 'ReadingOrder')
region_order_sub = ET.SubElement(region_order, 'OrderedGroup')
region_order_sub.set('id',"ro357564684568544579089")
region_order=ET.SubElement(page, 'ReadingOrder') args_sort=np.argsort(order_of_texts)
region_order_sub = ET.SubElement(region_order, 'OrderedGroup') for vj in args_sort:
name="coord_text_"+str(vj)
name = ET.SubElement(region_order_sub, 'RegionRefIndexed')
name.set('index',str(order_of_texts[vj]) )
name.set('regionRef',id_of_texts[vj])
region_order_sub.set('id',"ro357564684568544579089")
args_sort=np.argsort(order_of_texts) id_indexer=0
for vj in args_sort: id_indexer_l=0
name="coord_text_"+str(vj)
name = ET.SubElement(region_order_sub, 'RegionRefIndexed')
name.set('index',str(order_of_texts[vj]) )
name.set('regionRef',id_of_texts[vj])
for mm in range(len(found_polygons_text_region)):
textregion=ET.SubElement(page, 'TextRegion')
id_indexer=0 textregion.set('id','r'+str(id_indexer))
id_indexer_l=0 id_indexer+=1
for mm in range(len(found_polygons_text_region)): textregion.set('type','paragraph')
textregion=ET.SubElement(page, 'TextRegion') #if mm==0:
# textregion.set('type','heading')
#else:
# textregion.set('type','paragraph')
coord_text = ET.SubElement(textregion, 'Coords')
textregion.set('id','r'+str(id_indexer)) points_co=''
id_indexer+=1 for lmm in range(len(found_polygons_text_region[mm])):
if len(found_polygons_text_region[mm][lmm])==2:
points_co=points_co+str( int( (found_polygons_text_region[mm][lmm][0] +page_coord[2])/self.scale_x ) )
points_co=points_co+','
points_co=points_co+str( int( (found_polygons_text_region[mm][lmm][1] +page_coord[0])/self.scale_y ) )
else:
points_co=points_co+str( int((found_polygons_text_region[mm][lmm][0][0] +page_coord[2])/self.scale_x) )
points_co=points_co+','
points_co=points_co+str( int((found_polygons_text_region[mm][lmm][0][1] +page_coord[0])/self.scale_y) )
textregion.set('type','paragraph') if lmm<(len(found_polygons_text_region[mm])-1):
#if mm==0: points_co=points_co+' '
# textregion.set('type','heading') #print(points_co)
#else: coord_text.set('points',points_co)
# textregion.set('type','paragraph')
coord_text = ET.SubElement(textregion, 'Coords')
points_co=''
for lmm in range(len(found_polygons_text_region[mm])):
if len(found_polygons_text_region[mm][lmm])==2:
points_co=points_co+str( int( (found_polygons_text_region[mm][lmm][0] +page_coord[2])/self.scale_x ) )
points_co=points_co+','
points_co=points_co+str( int( (found_polygons_text_region[mm][lmm][1] +page_coord[0])/self.scale_y ) )
else:
points_co=points_co+str( int((found_polygons_text_region[mm][lmm][0][0] +page_coord[2])/self.scale_x) )
points_co=points_co+','
points_co=points_co+str( int((found_polygons_text_region[mm][lmm][0][1] +page_coord[0])/self.scale_y) )
if lmm<(len(found_polygons_text_region[mm])-1):
points_co=points_co+' '
#print(points_co)
coord_text.set('points',points_co)
for j in range(len(self.all_found_texline_polygons[mm])):
textline=ET.SubElement(textregion, 'TextLine')
for j in range(len(self.all_found_texline_polygons[mm])): textline.set('id','l'+str(id_indexer_l))
textline=ET.SubElement(textregion, 'TextLine') id_indexer_l+=1
textline.set('id','l'+str(id_indexer_l))
id_indexer_l+=1 coord = ET.SubElement(textline, 'Coords')
texteq=ET.SubElement(textline, 'TextEquiv')
coord = ET.SubElement(textline, 'Coords') uni=ET.SubElement(texteq, 'Unicode')
uni.text = ' '
texteq=ET.SubElement(textline, 'TextEquiv') #points = ET.SubElement(coord, 'Points')
uni=ET.SubElement(texteq, 'Unicode') points_co=''
uni.text = ' ' for l in range(len(self.all_found_texline_polygons[mm][j])):
#point = ET.SubElement(coord, 'Point')
#points = ET.SubElement(coord, 'Points')
points_co=''
for l in range(len(self.all_found_texline_polygons[mm][j])):
#point = ET.SubElement(coord, 'Point')
#point.set('x',str(found_polygons[j][l][0]))
#point.set('y',str(found_polygons[j][l][1]))
if len(self.all_found_texline_polygons[mm][j][l])==2:
points_co=points_co+str( int( (self.all_found_texline_polygons[mm][j][l][0] +page_coord[2]
+self.all_box_coord[mm][2])/self.scale_x) )
points_co=points_co+','
points_co=points_co+str( int( (self.all_found_texline_polygons[mm][j][l][1] +page_coord[0]
+self.all_box_coord[mm][0])/self.scale_y) )
else:
points_co=points_co+str( int( ( self.all_found_texline_polygons[mm][j][l][0][0] +page_coord[2]
+self.all_box_coord[mm][2])/self.scale_x ) )
points_co=points_co+','
points_co=points_co+str( int( ( self.all_found_texline_polygons[mm][j][l][0][1] +page_coord[0]
+self.all_box_coord[mm][0])/self.scale_y) )
#point.set('x',str(found_polygons[j][l][0])) if l<(len(self.all_found_texline_polygons[mm][j])-1):
#point.set('y',str(found_polygons[j][l][1])) points_co=points_co+' '
if len(self.all_found_texline_polygons[mm][j][l])==2: #print(points_co)
points_co=points_co+str( int( (self.all_found_texline_polygons[mm][j][l][0] +page_coord[2] coord.set('points',points_co)
+self.all_box_coord[mm][2])/self.scale_x) )
points_co=points_co+','
points_co=points_co+str( int( (self.all_found_texline_polygons[mm][j][l][1] +page_coord[0]
+self.all_box_coord[mm][0])/self.scale_y) )
else:
points_co=points_co+str( int( ( self.all_found_texline_polygons[mm][j][l][0][0] +page_coord[2]
+self.all_box_coord[mm][2])/self.scale_x ) )
points_co=points_co+','
points_co=points_co+str( int( ( self.all_found_texline_polygons[mm][j][l][0][1] +page_coord[0]
+self.all_box_coord[mm][0])/self.scale_y) )
if l<(len(self.all_found_texline_polygons[mm][j])-1):
points_co=points_co+' '
#print(points_co)
coord.set('points',points_co)
texteqreg=ET.SubElement(textregion, 'TextEquiv') texteqreg=ET.SubElement(textregion, 'TextEquiv')
unireg=ET.SubElement(texteqreg, 'Unicode') unireg=ET.SubElement(texteqreg, 'Unicode')
unireg.text = ' ' unireg.text = ' '
@ -1596,15 +1595,21 @@ class textlineerkenner:
image_page,page_coord=self.extract_page() image_page,page_coord=self.extract_page()
text_regions=self.extract_text_regions(image_page) text_regions=self.extract_text_regions(image_page)
boxes,contours=self.get_text_region_contours_and_boxes(text_regions) boxes,contours=self.get_text_region_contours_and_boxes(text_regions)
self.get_all_image_patches_based_on_text_regions(boxes,image_page)
textline_mask_tot=self.textline_contours(image_page)
indexes_sorted, matrix_of_orders=self.order_of_regions(textline_mask_tot,contours) if len(contours)>0:
order_of_texts, id_of_texts=self.order_and_id_of_texts(contours ,matrix_of_orders ,indexes_sorted ) self.get_all_image_patches_based_on_text_regions(boxes,image_page)
textline_mask_tot=self.textline_contours(image_page)
indexes_sorted, matrix_of_orders=self.order_of_regions(textline_mask_tot,contours)
order_of_texts, id_of_texts=self.order_and_id_of_texts(contours ,matrix_of_orders ,indexes_sorted )
self.get_textlines_for_each_textregions(textline_mask_tot,boxes) self.get_textlines_for_each_textregions(textline_mask_tot,boxes)
self.get_slopes_for_each_text_region(contours) self.get_slopes_for_each_text_region(contours)
self.deskew_textline_patches(contours, boxes) self.deskew_textline_patches(contours, boxes)
else:
contours=[]
order_of_texts=None
id_of_texts=None
self.write_into_page_xml(contours,page_coord,self.dir_out , order_of_texts , id_of_texts) self.write_into_page_xml(contours,page_coord,self.dir_out , order_of_texts , id_of_texts)
# Destroy the current Keras session/graph to free memory # Destroy the current Keras session/graph to free memory

Write Preview
Loading…
Cancel
Save