qurator-spk/sbb_textline_detection
1
0
Fork 0
mirror of https://github.com/qurator-spk/sbb_textline_detection.git synced 2025-10-31 17:34:16 +01:00
Code Issues Releases Wiki Activity

resolving issue https://github.com/qurator-spk/sbb_textline_detection/issues/53

Browse source
This commit is contained in:
vahidrezanezhad 2020-12-22 21:29:34 +01:00 • committed by GitHub
parent 17069f72d2
commit 4c498fcad2
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 82 additions and 75 deletions
Download patch file Download diff file Expand all files Collapse all files

145
qurator/sbb_textline_detector/main.py
View file

@ -2066,89 +2066,96 @@ class textline_detector:
gc.collect() gc.collect()
t2=time.time() t2=time.time()
try: try:
# extract text regions and corresponding contours and surrounding box try:
text_regions=self.extract_text_regions(image_page) # extract text regions and corresponding contours and surrounding box
text_regions=self.extract_text_regions(image_page)
text_regions = cv2.erode(text_regions, self.kernel, iterations=3) text_regions = cv2.erode(text_regions, self.kernel, iterations=3)
text_regions = cv2.dilate(text_regions, self.kernel, iterations=4) text_regions = cv2.dilate(text_regions, self.kernel, iterations=4)
#plt.imshow(text_regions[:,:,0]) #plt.imshow(text_regions[:,:,0])
#plt.show() #plt.show()
contours=self.get_text_region_contours_and_boxes(text_regions) contours=self.get_text_region_contours_and_boxes(text_regions)
########## ##########
K.clear_session()
gc.collect()
except:
text_regions=None
contours=[]
t3=time.time()
if len(contours)>0:
# extracting textlines using segmentation
textline_mask_tot=self.textline_contours(image_page)
##########
K.clear_session()
gc.collect()
t4=time.time()
# calculate the slope for deskewing for each box of text region.
contours=self.get_slopes_and_deskew(contours,textline_mask_tot)
gc.collect()
t5=time.time()
# get orders of each textregion. This method by now only works for one column documents.
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 )
##########
gc.collect()
t6=time.time()
self.get_all_image_patches_coordination(image_page)
##########
##########
gc.collect()
t7=time.time()
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)
# Destroy the current Keras session/graph to free memory
K.clear_session() K.clear_session()
gc.collect()
print( "time total = "+"{0:.2f}".format(time.time()-t1) )
print( "time needed for page extraction = "+"{0:.2f}".format(t2-t1) )
print( "time needed for text region extraction and get contours = "+"{0:.2f}".format(t3-t2) )
if len(contours)>0:
print( "time needed for textlines = "+"{0:.2f}".format(t4-t3) )
print( "time needed to get slopes of regions (deskewing) = "+"{0:.2f}".format(t5-t4) )
print( "time needed to get order of regions = "+"{0:.2f}".format(t6-t5) )
print( "time needed to implement deskewing = "+"{0:.2f}".format(t7-t6) )
except: except:
text_regions=None
contours=[]
t3=time.time()
if len(contours)>0:
# extracting textlines using segmentation
textline_mask_tot=self.textline_contours(image_page)
##########
K.clear_session()
gc.collect()
t4=time.time()
# calculate the slope for deskewing for each box of text region.
contours=self.get_slopes_and_deskew(contours,textline_mask_tot)
gc.collect()
t5=time.time()
# get orders of each textregion. This method by now only works for one column documents.
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 )
##########
gc.collect()
t6=time.time()
self.get_all_image_patches_coordination(image_page)
##########
##########
gc.collect()
t7=time.time()
else:
contours=[] contours=[]
order_of_texts=None order_of_texts=None
id_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)
print( "time total = "+"{0:.2f}".format(time.time()-t1) )
# Destroy the current Keras session/graph to free memory
K.clear_session()
print( "time total = "+"{0:.2f}".format(time.time()-t1) )
print( "time needed for page extraction = "+"{0:.2f}".format(t2-t1) )
print( "time needed for text region extraction and get contours = "+"{0:.2f}".format(t3-t2) )
if len(contours)>0:
print( "time needed for textlines = "+"{0:.2f}".format(t4-t3) )
print( "time needed to get slopes of regions (deskewing) = "+"{0:.2f}".format(t5-t4) )
print( "time needed to get order of regions = "+"{0:.2f}".format(t6-t5) )
print( "time needed to implement deskewing = "+"{0:.2f}".format(t7-t6) )