def estimate_angle(raw, maxskew=2, skewsteps=8, perc=80, range=20, zoom=0.5, bignore=0.1):
comment = ""
rawF = read_image_gray(raw)
# perform image normalization
image = rawF - amin(rawF)
if amax(image) == amin(image):
print "# image is empty", fname
return
image /= amax(image)
extreme = (sum(image < 0.05) + sum(image > 0.95)) * 1.0 / prod(image.shape)
if extreme > 0.95:
comment += " no-normalization"
flat = image
else:
# check whether the image is already effectively binarized
# if not, we need to flatten it by estimating the local whitelevel
m = interpolation.zoom(image, zoom)
m = filters.percentile_filter(m, perc, size=(range, 2))
m = filters.percentile_filter(m, perc, size=(2, range))
m = interpolation.zoom(m, 1.0 / zoom)
w, h = minimum(array(image.shape), array(m.shape))
flat = clip(image[:w, :h] - m[:w, :h] + 1, 0, 1)
# estimate skew angle and rotate
d0, d1 = flat.shape
o0, o1 = int(bignore * d0), int(bignore * d1)
flat = amax(flat) - flat
flat -= amin(flat)
est = flat[o0 : d0 - o0, o1 : d1 - o1]
ma = maxskew
ms = int(2 * maxskew * skewsteps)
angle = estimate_skew_angle(est, linspace(-ma, ma, ms + 1))
return angle
def nlbin(im, threshold=0.5, zoom=0.5, escale=1.0, border=0.1, perc=80,
range=20, low=5, high=90):
"""
Performs binarization using non-linear processing.
Args:
im (PIL.Image):
threshold (float):
zoom (float): Zoom for background page estimation
escale (float): Scale for estimating a mask over the text region
border (float): Ignore this much of the border
perc (int): Percentage for filters
range (int): Range for filters
low (int): Percentile for black estimation
high (int): Percentile for white estimation
Returns:
PIL.Image containing the binarized image
"""
if im.mode == '1':
return im
raw = pil2array(im)
# rescale image to between -1 or 0 and 1
raw = raw/np.float(np.iinfo(raw.dtype).max)
if raw.ndim == 3:
raw = np.mean(raw, 2)
# perform image normalization
if np.amax(raw) == np.amin(raw):
raise KrakenInputException('Image is empty')
image = raw-np.amin(raw)
image /= np.amax(image)
m = interpolation.zoom(image, zoom)
m = filters.percentile_filter(m, perc, size=(range, 2))
m = filters.percentile_filter(m, perc, size=(2, range))
m = interpolation.zoom(m, 1.0/zoom)
w, h = np.minimum(np.array(image.shape), np.array(m.shape))
flat = np.clip(image[:w, :h]-m[:w, :h]+1, 0, 1)
# estimate low and high thresholds
d0, d1 = flat.shape
o0, o1 = int(border*d0), int(border*d1)
est = flat[o0:d0-o0, o1:d1-o1]
# by default, we use only regions that contain
# significant variance; this makes the percentile
# based low and high estimates more reliable
v = est-filters.gaussian_filter(est, escale*20.0)
v = filters.gaussian_filter(v**2, escale*20.0)**0.5
v = (v > 0.3*np.amax(v))
v = morphology.binary_dilation(v, structure=np.ones((escale*50, 1)))
v = morphology.binary_dilation(v, structure=np.ones((1, escale*50)))
est = est[v]
lo = np.percentile(est.ravel(), low)
hi = np.percentile(est.ravel(), high)
flat -= lo
flat /= (hi-lo)
flat = np.clip(flat, 0, 1)
bin = np.array(255*(flat > threshold), 'B')
return array2pil(bin)
def cleanMe(spec, window=75, p=10):
# returns a cleaned spectrum, removing extreme outlier peaks and troughs
a = np.array(spec)
topcut = percentile_filter( a, 100-p, size=window )
a[ a>topcut ] = topcut[ a>topcut ]
botcut = percentile_filter( a, p, size=window )
a[ a<botcut ] = botcut[ a<botcut ]
return a
def order_statistics_filter(grid,window_size=(3,3,3),statistics_type='median',rank=1):
filtered = grid.copy()
if statistics_type=='minimum':
scifilt.minimum_filter(grid,window_size,None,filtered, mode='nearest')
elif statistics_type=='maximum':
scifilt.maximum_filter(grid,window_size,None,filtered, mode='nearest')
elif statistics_type=='median':
scifilt.median_filter(grid,window_size,None,filtered, mode='nearest')
elif statistics_type[:-2]=='percentile' or statistics_type[:-2]=='per':
per = np.int(statistics_type[-2:])
scifilt.percentile_filter(grid,per,window_size,None,filtered, mode='nearest')
elif statistics_type=='rank':
scifilt.rank_filter(grid,rank,window_size,None,filtered, mode='nearest')
return filtered
return filtered
def normalize(self, method='percentile', window=None, perc=20, offset=0.1):
"""
Normalize by subtracting and dividing by a baseline.
Baseline can be derived from a global mean or percentile,
or a smoothed percentile estimated within a rolling window.
Windowed baselines may only be well-defined for
temporal series data.
Parameters
----------
baseline : str, optional, default = 'percentile'
Quantity to use as the baseline, options are 'mean', 'percentile',
'window', or 'window-exact'.
window : int, optional, default = 6
Size of window for baseline estimation,
for 'window' and 'window-exact' baseline only.
perc : int, optional, default = 20
Percentile value to use, for 'percentile',
'window', or 'window-exact' baseline only.
offset : float, optional, default = 0.1
Scalar added to baseline during division to avoid division by 0.
"""
check_options(method, ['mean', 'percentile', 'window', 'window-exact'])
from warnings import warn
if not (method == 'window' or method == 'window-exact') and window is not None:
warn('Setting window without using method "window" has no effect')
if method == 'mean':
baseFunc = mean
if method == 'percentile':
baseFunc = lambda x: percentile(x, perc)
if method == 'window':
from scipy.ndimage.filters import percentile_filter
baseFunc = lambda x: percentile_filter(x.astype(float64), perc, window, mode='nearest')
if method == 'window-exact':
if window & 0x1:
left, right = (ceil(window/2), ceil(window/2) + 1)
else:
left, right = (window/2, window/2)
n = len(self.index)
baseFunc = lambda x: asarray([percentile(x[max(ix-left, 0):min(ix+right+1, n)], perc)
for ix in arange(0, n)])
def get(y):
b = baseFunc(y)
return (y - b) / (b + offset)
return self.map(get)
def normalize(self, baseline='percentile', window=None, perc=20):
"""
Normalize each time series by subtracting and dividing by a baseline.
Baseline can be derived from a global mean or percentile,
or a smoothed percentile estimated within a rolling window.
Parameters
----------
baseline : str, optional, default = 'percentile'
Quantity to use as the baseline, options are 'mean', 'percentile', 'window', or 'window-fast'
window : int, optional, default = 6
Size of window for baseline estimation, for 'window' and 'window-fast' baseline only
perc : int, optional, default = 20
Percentile value to use, for 'percentile', 'window', or 'window-fast' baseline only
"""
checkParams(baseline, ['mean', 'percentile', 'window', 'window-fast'])
method = baseline.lower()
from warnings import warn
if not (method == 'window' or method == 'window-fast') and window is not None:
warn('Setting window without using method "window" has no effect')
if method == 'mean':
baseFunc = mean
if method == 'percentile':
baseFunc = lambda x: percentile(x, perc)
if method == 'window':
if window & 0x1:
left, right = (ceil(window/2), ceil(window/2) + 1)
else:
left, right = (window/2, window/2)
n = len(self.index)
baseFunc = lambda x: asarray([percentile(x[max(ix-left, 0):min(ix+right+1, n)], perc)
for ix in arange(0, n)])
if method == 'window-fast':
from scipy.ndimage.filters import percentile_filter
baseFunc = lambda x: percentile_filter(x.astype(float64), perc, window, mode='nearest')
def get(y):
b = baseFunc(y)
return (y - b) / (b + 0.1)
return self.applyValues(get, keepIndex=True)
def deadPixelMask(self, pic):
'''
pixels with much lower intensity compare to adjacent pixels will be masked
:param pic: 2d array, image array to be processed
:return: 2d array of boolean, 1 stands for masked pixel
'''
avgpic = np.average(pic)
ks = np.ones((5, 5))
ks1 = np.ones((7, 7))
picb = snf.percentile_filter(pic, 5, 3) < avgpic / 10
picb = snm.binary_dilation(picb, structure=ks)
picb = snm.binary_erosion(picb, structure=ks1)
return picb
def darkPixelMask(self, pic, r=None):
'''
pixels with much lower intensity compare to adjacent pixels will be masked
:param pic: 2d array, image array to be processed
:param r: float, a threshold for masked pixels
:return: 2d array of boolean, 1 stands for masked pixel
'''
r = self.config.darkpixelr if r == None else r # 0.1
avgpic = np.average(pic)
ks = np.ones((5, 5))
ks1 = np.ones((7, 7))
picb = snf.percentile_filter(pic, 5, 3) < avgpic * r
picb = snm.binary_dilation(picb, structure=ks)
picb = snm.binary_erosion(picb, structure=ks1)
return picb
def ndi_med(image, n):
return percentile_filter(image, 50, size=n * 2 - 1)
plt.xticks(range(0, 1500, 300), [0, '', '', 30, ''])
plt.xlim(0, 1200)
plt.ylim(0, 1.)
plt.xlabel('Time [s]', labelpad=-10)
plt.ylabel('Activity')
# real data
try:
data = loadmat(filename)
fmean_roi = data['obj']['timeSeriesArrayHash'].item()[0]['value'].item()[0][
0]['valueMatrix'].item().ravel()
fmean_neuropil = data['obj']['timeSeriesArrayHash'].item()[0]['value'].item()[0][
1]['valueMatrix'].item().ravel()
fmean_comp = np.ravel(fmean_roi - 0.7 * fmean_neuropil)
b = percentile_filter(fmean_comp, 20, 3000, mode='nearest')
mu = .075
# shift such that baseline is not negative, but ~0
y = ((fmean_comp - b) / (b + 10)).astype(float) + mu
t_frame = data['obj']['timeSeriesArrayHash'].item()[0]['value'].item()[0][
0]['time'].item().ravel()
filt = data['obj']['timeSeriesArrayHash'].item()[0]['value'].item()[0][3][
'valueMatrix'].item().ravel()
t_ephys = data['obj']['timeSeriesArrayHash'].item()[0]['value'].item()[0][
3]['time'].item().ravel()
detected_spikes = data['obj']['timeSeriesArrayHash'].item()[0]['value'].item()[0][4][
'valueMatrix'].item().ravel()
spike_time = t_ephys[detected_spikes.astype(bool)]
def nlbin(im: Image,
threshold: float = 0.5,
zoom: float = 0.5,
escale: float = 1.0,
border: float = 0.1,
perc: int = 80,
range: int = 20,
low: int = 5,
high: int = 90) -> Image:
"""
Performs binarization using non-linear processing.
Args:
im (PIL.Image):
threshold (float):
zoom (float): Zoom for background page estimation
escale (float): Scale for estimating a mask over the text region
border (float): Ignore this much of the border
perc (int): Percentage for filters
range (int): Range for filters
low (int): Percentile for black estimation
high (int): Percentile for white estimation
Returns:
PIL.Image containing the binarized image
Raises:
KrakenInputException when trying to binarize an empty image.
"""
im_str = get_im_str(im)
logger.info('Binarizing {}'.format(im_str))
if is_bitonal(im):
logger.info(
'Skipping binarization because {} is bitonal.'.format(im_str))
return im
# convert to grayscale first
logger.debug('Converting {} to grayscale'.format(im_str))
im = im.convert('L')
raw = pil2array(im)
logger.debug('Scaling and normalizing')
# rescale image to between -1 or 0 and 1
raw = raw / np.float(np.iinfo(raw.dtype).max)
# perform image normalization
if np.amax(raw) == np.amin(raw):
logger.warning('Trying to binarize empty image {}'.format(im_str))
raise KrakenInputException('Image is empty')
image = raw - np.amin(raw)
image /= np.amax(image)
logger.debug('Interpolation and percentile filtering')
with warnings.catch_warnings():
warnings.simplefilter('ignore', UserWarning)
m = interpolation.zoom(image, zoom)
m = filters.percentile_filter(m, perc, size=(range, 2))
m = filters.percentile_filter(m, perc, size=(2, range))
mh, mw = m.shape
oh, ow = image.shape
scale = np.diag([mh * 1.0 / oh, mw * 1.0 / ow])
m = affine_transform(m, scale, output_shape=image.shape)
w, h = np.minimum(np.array(image.shape), np.array(m.shape))
flat = np.clip(image[:w, :h] - m[:w, :h] + 1, 0, 1)
# estimate low and high thresholds
d0, d1 = flat.shape
o0, o1 = int(border * d0), int(border * d1)
est = flat[o0:d0 - o0, o1:d1 - o1]
logger.debug('Threshold estimates {}'.format(est))
# by default, we use only regions that contain
# significant variance; this makes the percentile
# based low and high estimates more reliable
logger.debug('Refine estimates')
v = est - filters.gaussian_filter(est, escale * 20.0)
v = filters.gaussian_filter(v**2, escale * 20.0)**0.5
v = (v > 0.3 * np.amax(v))
v = morphology.binary_dilation(v, structure=np.ones((int(escale * 50), 1)))
v = morphology.binary_dilation(v, structure=np.ones((1, int(escale * 50))))
est = est[v]
lo = np.percentile(est.ravel(), low)
hi = np.percentile(est.ravel(), high)
flat -= lo
flat /= (hi - lo)
flat = np.clip(flat, 0, 1)
logger.debug('Thresholding at {}'.format(threshold))
bin = np.array(255 * (flat > threshold), 'B')
return array2pil(bin)
def ndi_med(image, n):
return percentile_filter(image, 50, size=n * 2 - 1)
def moving_window_atribute(data,atribute='mean',window=(3,3,3),percentile=50,rank=1,clip_limits=(0,1),fisher=True,border_mode='nearest'):
#reflect’, ‘constant’, ‘nearest’, ‘mirror’, ‘wrap’
# minimum,maximum,median,percentile,rank,mean,variance,std,clip,sum,product,peak2peak,signal2noise,skewness
# kurtosis
if atribute == 'minimum':
atrib = data.copy()
scifilt.minimum_filter(data,window,None,atrib, mode=border_mode)
return atrib
elif atribute == 'maximum':
atrib = data.copy()
scifilt.maximum_filter(data,window,None,atrib, mode=border_mode)
return atrib
elif atribute == 'median':
atrib = data.copy()
scifilt.median_filter(data,window,None,atrib, mode=border_mode)
return atrib
elif atribute == 'percentile':
atrib = data.copy()
scifilt.percentile_filter(data,percentile,window,None,atrib, mode=border_mode)
return atrib
elif atribute == 'rank':
atrib = data.copy()
scifilt.rank_filter(data,rank,window,None,atrib, mode=border_mode)
return atrib
elif atribute == 'mean':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
atrib[i,j,k] = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].mean()
return atrib
elif atribute == 'variance':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
atrib[i,j,k] = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].var()
return atrib
elif atribute == 'std':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
atrib[i,j,k] = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].std()
return atrib
elif atribute == 'clip':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
m = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].flatten()
l0 = np.percentile(m,clip_limits[0])
l1 = np.percentile(m,clip_limits[1])
m.clip(l0,l1)
atrib[i,j,k] = np.percentile(m,50)
return atrib
elif atribute == 'sum':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
atrib[i,j,k] = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].sum()
return atrib
elif atribute == 'product':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
atrib[i,j,k] = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].prod()
return atrib
elif atribute == 'peak2peak':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
atrib[i,j,k] = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].ptp()
return atrib
elif atribute == 'signal2noise':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
m = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].mean()
v = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].std()
atrib[i,j,k] = m/v
return atrib
elif atribute == 'skewness':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
m = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].flatten()
v = st.skew(m)
atrib[i,j,k] = v
return atrib
elif atribute == 'kurtosis':
atrib = data.copy()
blocks = atrib.shape
for i in xrange(blocks[0]):
for j in xrange(blocks[1]):
for k in xrange(blocks[2]):
m = data[np.clip(i-window[0],0,blocks[0]):i+window[0]+1,np.clip(j-window[1],0,blocks[1]):j+window[1]+1,np.clip(k-window[2],0,blocks[2]):k+window[2]+1].flatten()
v = st.kurtosis(m,fisher=fisher)
atrib[i,j,k] = v
return atrib
else:
return False
def low_high(d, winsize):
high = percentile_filter(d, 95, winsize)
low = percentile_filter(d, 5, winsize)
high = filters.lowpass_fir(high, winsize)
low = filters.lowpass_fir(low, winsize)
return low, high
def _process_segment(self, page_image, page, page_xywh, page_id,
input_file, n):
LOG = getLogger('OcrdAnybaseocrBinarizer')
raw = ocrolib.pil2array(page_image)
if len(raw.shape) > 2:
raw = np.mean(raw, 2)
raw = raw.astype("float64")
# perform image normalization
image = raw - amin(raw)
if amax(image) == amin(image):
LOG.info("# image is empty: %s" % (page_id))
return
image /= amax(image)
# check whether the image is already effectively binarized
if self.parameter['gray']:
extreme = 0
else:
extreme = (np.sum(image < 0.05) +
np.sum(image > 0.95)) * 1.0 / np.prod(image.shape)
if extreme > 0.95:
comment = "no-normalization"
flat = image
else:
comment = ""
# if not, we need to flatten it by estimating the local whitelevel
LOG.info("Flattening")
m = interpolation.zoom(image, self.parameter['zoom'])
m = filters.percentile_filter(m,
self.parameter['perc'],
size=(self.parameter['range'], 2))
m = filters.percentile_filter(m,
self.parameter['perc'],
size=(2, self.parameter['range']))
m = interpolation.zoom(m, 1.0 / self.parameter['zoom'])
if self.parameter['debug'] > 0:
clf()
imshow(m, vmin=0, vmax=1)
ginput(1, self.parameter['debug'])
w, h = minimum(array(image.shape), array(m.shape))
flat = clip(image[:w, :h] - m[:w, :h] + 1, 0, 1)
if self.parameter['debug'] > 0:
clf()
imshow(flat, vmin=0, vmax=1)
ginput(1, self.parameter['debug'])
# estimate low and high thresholds
LOG.info("Estimating Thresholds")
d0, d1 = flat.shape
o0, o1 = int(self.parameter['bignore'] * d0), int(
self.parameter['bignore'] * d1)
est = flat[o0:d0 - o0, o1:d1 - o1]
if self.parameter['escale'] > 0:
# by default, we use only regions that contain
# significant variance; this makes the percentile
# based low and high estimates more reliable
e = self.parameter['escale']
v = est - filters.gaussian_filter(est, e * 20.0)
v = filters.gaussian_filter(v**2, e * 20.0)**0.5
v = (v > 0.3 * amax(v))
v = morphology.binary_dilation(v, structure=ones((int(e * 50), 1)))
v = morphology.binary_dilation(v, structure=ones((1, int(e * 50))))
if self.parameter['debug'] > 0:
imshow(v)
ginput(1, self.parameter['debug'])
est = est[v]
lo = stats.scoreatpercentile(est.ravel(), self.parameter['lo'])
hi = stats.scoreatpercentile(est.ravel(), self.parameter['hi'])
# rescale the image to get the gray scale image
LOG.info("Rescaling")
flat -= lo
flat /= (hi - lo)
flat = clip(flat, 0, 1)
if self.parameter['debug'] > 0:
imshow(flat, vmin=0, vmax=1)
ginput(1, self.parameter['debug'])
binarized = 1 * (flat > self.parameter['threshold'])
# output the normalized grayscale and the thresholded images
# print_info("%s lo-hi (%.2f %.2f) angle %4.1f %s" % (fname, lo, hi, angle, comment))
LOG.info("%s lo-hi (%.2f %.2f) %s" % (page_id, lo, hi, comment))
LOG.info("writing")
if self.parameter['debug'] > 0 or self.parameter['show']:
clf()
gray()
imshow(binarized)
ginput(1, max(0.1, self.parameter['debug']))
page_xywh['features'] += ',binarized'
bin_array = array(255 * (binarized > ocrolib.midrange(binarized)), 'B')
bin_image = ocrolib.array2pil(bin_array)
file_id = make_file_id(input_file, self.output_file_grp)
file_path = self.workspace.save_image_file(
bin_image,
file_id + '-IMG',
page_id=page_id,
file_grp=self.output_file_grp)
page.add_AlternativeImage(
AlternativeImageType(filename=file_path,
comments=page_xywh['features']))
def dbl_pct_filt(arr):
# Define percentile filter for clipping off artefactual IAS protrusions due to dangling epidermis
out = percentile_filter(percentile_filter(arr, size=30, percentile=10),
size=30,
percentile=90)
return out
def nlbin(im: Image.Image,
threshold: float = 0.5,
zoom: float = 0.5,
escale: float = 1.0,
border: float = 0.1,
perc: int = 80,
range: int = 20,
low: int = 5,
high: int = 90) -> Image:
"""
Performs binarization using non-linear processing.
Args:
im (PIL.Image.Image):
threshold (float):
zoom (float): Zoom for background page estimation
escale (float): Scale for estimating a mask over the text region
border (float): Ignore this much of the border
perc (int): Percentage for filters
range (int): Range for filters
low (int): Percentile for black estimation
high (int): Percentile for white estimation
Returns:
PIL.Image containing the binarized image
Raises:
KrakenInputException when trying to binarize an empty image.
"""
im_str = get_im_str(im)
logger.info('Binarizing {}'.format(im_str))
if is_bitonal(im):
logger.info('Skipping binarization because {} is bitonal.'.format(im_str))
return im
# convert to grayscale first
logger.debug('Converting {} to grayscale'.format(im_str))
im = im.convert('L')
raw = pil2array(im)
logger.debug('Scaling and normalizing')
# rescale image to between -1 or 0 and 1
raw = raw/np.float(np.iinfo(raw.dtype).max)
# perform image normalization
if np.amax(raw) == np.amin(raw):
logger.warning('Trying to binarize empty image {}'.format(im_str))
raise KrakenInputException('Image is empty')
image = raw-np.amin(raw)
image /= np.amax(image)
logger.debug('Interpolation and percentile filtering')
with warnings.catch_warnings():
warnings.simplefilter('ignore', UserWarning)
m = interpolation.zoom(image, zoom)
m = filters.percentile_filter(m, perc, size=(range, 2))
m = filters.percentile_filter(m, perc, size=(2, range))
mh, mw = m.shape
oh, ow = image.shape
scale = np.diag([mh * 1.0/oh, mw * 1.0/ow])
m = affine_transform(m, scale, output_shape=image.shape)
w, h = np.minimum(np.array(image.shape), np.array(m.shape))
flat = np.clip(image[:w, :h]-m[:w, :h]+1, 0, 1)
# estimate low and high thresholds
d0, d1 = flat.shape
o0, o1 = int(border*d0), int(border*d1)
est = flat[o0:d0-o0, o1:d1-o1]
logger.debug('Threshold estimates {}'.format(est))
# by default, we use only regions that contain
# significant variance; this makes the percentile
# based low and high estimates more reliable
logger.debug('Refine estimates')
v = est-filters.gaussian_filter(est, escale*20.0)
v = filters.gaussian_filter(v**2, escale*20.0)**0.5
v = (v > 0.3*np.amax(v))
v = morphology.binary_dilation(v, structure=np.ones((int(escale * 50), 1)))
v = morphology.binary_dilation(v, structure=np.ones((1, int(escale * 50))))
est = est[v]
lo = np.percentile(est.ravel(), low)
hi = np.percentile(est.ravel(), high)
flat -= lo
flat /= (hi-lo)
flat = np.clip(flat, 0, 1)
logger.debug('Thresholding at {}'.format(threshold))
bin = np.array(255*(flat > threshold), 'B')
return array2pil(bin)
