def test_threshold_sauvola_iterable_window_size(self):
ref = np.array([[False, False, False, True, True],
[False, False, True, True, True],
[False, False, True, True, False],
[False, True, True, True, False],
[True, True, False, False, False]])
thres = threshold_sauvola(self.image, window_size=(3, 5), k=0.2, r=128)
out = self.image > thres
assert_array_equal(ref, out)
def test_threshold_sauvola(self):
ref = np.array(
[[False, False, False, True, True],
[False, False, True, True, True],
[False, False, True, True, False],
[False, True, True, True, False],
[True, True, False, False, False]]
)
thres = threshold_sauvola(self.image, window_size=3, k=0.2, r=128)
out = self.image > thres
assert_equal(ref, out)
def deskew(args,image, image_param):
# Deskew
# Calculate the angle of the points between 20% and 80% of the line
uintimage = get_uintimg(image)
thresh = th.threshold_sauvola(uintimage, args.threshwindow, args.threshweight)
binary = uintimage > thresh
binary = 1-binary #inverse binary
binary = np.rot90(binary,args.horlinepos)
labels, numl = measurements.label(binary)
objects = measurements.find_objects(labels)
deskew_path = None
for i, b in enumerate(objects):
linecoords = Linecoords(image, i, b)
# The line has to be bigger than minwidth, smaller than maxwidth, stay in the top (30%) of the img,
# only one obj allowed and the line isn't allowed to start contact the topborder of the image
if int(args.minwidthhor * image_param.width) < get_width(b) < int(args.maxwidthhor * image_param.width) \
and int(image_param.height * args.minheighthor) < get_height(b) < int(image_param.height * args.maxheighthor) \
and int(image_param.height * args.minheighthormask) < (linecoords.height_start+linecoords.height_stop)/2 < int(image_param.height * args.maxheighthormask) \
and linecoords.height_start != 0:
pixelwidth = set_pixelground(binary[b].shape[1])
arr = np.arange(1, pixelwidth(args.deskewlinesize) + 1)
mean_y = []
#Calculate the mean value for every y-array
for idx in range(pixelwidth(args.deskewlinesize)):
value_y = measurements.find_objects(labels[b][:, idx + pixelwidth((1.0-args.deskewlinesize)/2)] == i + 1)[0]
mean_y.append((value_y[0].stop + value_y[0].start) / 2)
polyfit_value = np.polyfit(arr, mean_y, 1)
deskewangle = np.arctan(polyfit_value[0]) * (360 / (2 * np.pi))
args.ramp = True
deskew_image = transform.rotate(image, deskewangle)
deskew_path = "%s_deskew.%s" % (image_param.pathout+image_param.name, args.extension)
image_param.deskewpath = deskew_path
with warnings.catch_warnings():
#Transform rotate convert the img to float and save convert it back
warnings.simplefilter("ignore")
misc.imsave(deskew_path, deskew_image)
break
return deskew_path
def linecoords_analyse(args,origimg, image_param, clippingmask):
image = get_uintimg(origimg)
origimg = np.rot90(origimg, args.horlinepos)
thresh = th.threshold_sauvola(image, args.threshwindow, args.threshweight)
binary = image > thresh
binary = 1-binary #inverse binary
binary = np.rot90(binary, args.horlinepos)
labels, numl = measurements.label(binary)
objects = measurements.find_objects(labels)
count_height = 0
count_width = 0
pixelheight = set_pixelground(image_param.height)
pixelwidth = set_pixelground(image_param.width)
list_linecoords = [] # Init list of linecoordinates the format is: [0]: width.start, width.stopt,
# [1]:height.start, height.stop, [2]: Type of line [B = blank, L = vertical line]
for i, b in enumerate(objects):
# The line has to be bigger than minwidth, smaller than maxwidth, stay in the top (30%) of the img,
# only one obj allowed and the line isn't allowed to start contact the topborder of the image
linecoords = Linecoords(labels, i, b)
if pixelwidth(args.minwidthhor) < get_width(b) < pixelwidth(args.maxwidthhor) \
and pixelheight(args.minheighthor) < get_height(b) < pixelheight(args.maxheighthor) \
and pixelheight(args.minheighthormask) < linecoords.height_stop < pixelheight(args.maxheighthormask) \
and count_width == 0 \
and linecoords.height_start != 0:
# Distance Calculation - defining the clippingmask
border = get_mindist(b, image_param.width)
topline_width_stop = b[0].stop + 2 # Lowest Point of object + 2 Pixel
if clippingmask.user == None:
clippingmask.width_start = border
clippingmask.width_stop = image_param.width - border
#if clippingmask.width_start > pixelwidth(args.minwidthmask):
# clippingmask.width_start = pixelwidth(args.minwidthmask)
#if clippingmask.width_stop < pixelwidth(1.0-args.minwidthmask):
# clippingmask.width_stop = pixelwidth(1.0-args.minwidthmask)
clippingmask.height_start = copy.deepcopy(topline_width_stop)
clippingmask.height_stop = 0
# Test for cropping the area under the first vertical line
#roi = image[hobj_bottom:image_param.height, border:image_param.width - border] # region of interest
#imsave("%s_crop.%s" % (image_param.pathout+image_param.name, args.extension), roi)
# Get coordinats of the line
labels[b][labels[b] == i + 1] = 0
count_width += 1
if pixelheight(args.minheightver) < get_height(b) < pixelheight(args.maxheightver) \
and pixelwidth(args.minwidthver) < get_width(b) < pixelwidth(args.maxwidthver) \
and pixelwidth(args.minwidthvermask) < (linecoords.width_start+linecoords.width_stop)/2 < pixelwidth(args.maxwidthvermask) \
and float(get_width(b))/float(get_height(b)) < args.maxgradientver:
linecoords.segmenttype = 'L' # Defaultvalue for segmenttype 'P' for horizontal lines
if count_height == 0:
if b[0].start - topline_width_stop > pixelheight(args.minsizeblank+args.minsizeblankobolustop):
blankline = Linecoords(labels,i,b)
blankline.segmenttype = 'B'
blankline.height_start = topline_width_stop
blankline.height_stop = linecoords.height_start
blankline.width_start = border
blankline.width_stop = image_param.width - border
blankline.middle = int(((linecoords.width_start+linecoords.width_stop)-1)/2)
list_linecoords.append(copy.deepcopy(blankline))
count_height += 1
if args.ramp != None:
whiteout_ramp(origimg, linecoords)
list_linecoords.append(copy.deepcopy(linecoords))
count_height += 1
else:
# Should fix to short vertical lines, in the height to top if they appear before any B Part in the image
if topline_width_stop > 0:
linecoords.height_start = topline_width_stop + pixelheight(args.addstartheightab)
list_linecoords.append(copy.deepcopy(linecoords))
count_height += 1
if args.ramp != None:
whiteout_ramp(origimg, linecoords)
elif list_linecoords[count_height - 1].height_stop < b[0].stop:
#Test argument to filter braces
if b[0].start - list_linecoords[count_height - 1].height_stop > pixelheight(args.minsizeblank):
blankline = Linecoords(labels,i,b)
blankline.segmenttype = 'B'
blankline.height_start = list_linecoords[count_height - 1].height_stop
blankline.height_stop = linecoords.height_start
blankline.width_start = border
blankline.width_stop = image_param.width - border
blankline.middle = int(((linecoords.width_start+linecoords.width_stop)-1)/2)
list_linecoords.append(copy.deepcopy(blankline))
count_height += 1
list_linecoords.append(copy.deepcopy(linecoords))
if args.ramp != None:
whiteout_ramp(origimg, linecoords)
count_height += 1
labels[b][labels[b] == i + 1] = 0
else:
if args.ramp != None:
whiteout_ramp(origimg, linecoords)
print b[0].stop
list_linecoords[count_height - 1].height_stop = b[0].stop
labels[b][labels[b] == i + 1] = 0
#imsave("%s_EDIT%d.%s" % (image_param.pathout, linecoords.object_value, args.extension), image)
return list_linecoords, border, topline_width_stop
def get_binary(args, image):
thresh = th.threshold_sauvola(image, args.threshwindow, args.threshweight)
binary = image > thresh
binary = 1 - binary # inverse binary
binary = np.rot90(binary, args.horlinepos)
return binary
