def process(self, input):
"""
Crop an image, using IULIB. If any of
the parameters are -1 or less, use the
outer dimensions.
"""
x0, y0 = 0, 0
y1, x1 = input.shape
try:
x0 = int(self._params.get("x0", -1))
if x0 < 0: x0 = 0
except TypeError:
pass
try:
y0 = int(self._params.get("y0", -1))
if y0 < 0: y0 = 0
except TypeError:
pass
try:
x1 = int(self._params.get("x1", -1))
if x1 < 0: x1 = input.shape[1]
except TypeError:
pass
try:
y1 = int(self._params.get("y1", -1))
if y1 < 0: y1 = input.shape[0]
except TypeError:
pass
# flip the coord system from HOCR to internal
iy0 = input.shape[1] - y1
iy1 = input.shape[1] - y0i
iulibbin = ocrolib.numpy2narray(input)
out = ocrolib.iulib.bytearray()
ocrolib.iulib.extract_subimage(out, iulibbin, x0, iy0, x1, iy1)
return ocrolib.narray2numpy(out)
def process(self, binary, boxes):
"""Recognize page text.
input: tuple of binary, input boxes
return: page data
"""
self.prepare()
pageheight, pagewidth = binary.shape
iulibbin = ocrolib.numpy2narray(binary)
out = dict(bbox=[0, 0, pagewidth, pageheight], lines=[])
numlines = len(boxes.get("lines", []))
for i in range(numlines):
set_progress(self.logger, self.progress_func, i, numlines)
coords = boxes.get("lines")[i]
iulibcoords = (
coords[0], pageheight - coords[3], coords[2],
pageheight - coords[1])
lineimage = ocrolib.iulib.bytearray()
ocrolib.iulib.extract_subimage(lineimage, iulibbin, *iulibcoords)
out["lines"].append(dict(
index=i+1,
bbox=[coords[0], coords[1], coords[2], coords[3]],
text=self.get_transcript(ocrolib.narray2numpy(lineimage)),
))
set_progress(self.logger, self.progress_func, numlines, numlines)
self.cleanup()
return utils.hocr_from_data(out)
def process(self, input):
"""
Crop an image, using IULIB. If any of
the parameters are -1 or less, use the
outer dimensions.
"""
x0, y0 = 0, 0
y1, x1 = input.shape
try:
x0 = int(self._params.get("x0", -1))
if x0 < 0: x0 = 0
except TypeError: pass
try:
y0 = int(self._params.get("y0", -1))
if y0 < 0: y0 = 0
except TypeError: pass
try:
x1 = int(self._params.get("x1", -1))
if x1 < 0: x1 = input.shape[1]
except TypeError: pass
try:
y1 = int(self._params.get("y1", -1))
if y1 < 0: y1 = input.shape[0]
except TypeError: pass
# flip the coord system from HOCR to internal
iy0 = input.shape[1] - y1
iy1 = input.shape[1] - y0i
iulibbin = ocrolib.numpy2narray(input)
out = ocrolib.iulib.bytearray()
ocrolib.iulib.extract_subimage(out, iulibbin, x0, iy0, x1, iy1)
return ocrolib.narray2numpy(out)
def process(self, binary, boxes):
"""
Recognize page text.
input: tuple of binary, input boxes
return: page data
"""
self.prepare()
pageheight, pagewidth = binary.shape
iulibbin = ocrolib.numpy2narray(binary)
out = dict(bbox=[0, 0, pagewidth, pageheight], lines=[])
numlines = len(boxes.get("lines", []))
for i in range(numlines):
set_progress(self.logger, self.progress_func, i, numlines)
coords = boxes.get("lines")[i]
iulibcoords = (
coords[0], pageheight - coords[3], coords[2],
pageheight - coords[1])
lineimage = ocrolib.iulib.bytearray()
ocrolib.iulib.extract_subimage(lineimage, iulibbin, *iulibcoords)
out["lines"].append(dict(
index=i+1,
bbox=[coords[0], coords[1], coords[2], coords[3]],
text=self.get_transcript(ocrolib.narray2numpy(lineimage)),
))
set_progress(self.logger, self.progress_func, numlines, numlines)
self.cleanup()
return utils.hocr_from_data(out)
def process(self, input):
"""Blockout an image, using PIL. If any of
the parameters are -1 or less, use the
outer dimensions."""
height = input.shape[0]
pstr = self._params.get("boxes", "")
coords = get_coords(pstr)
if len(coords) == 0:
return input
sancoords = sanitise_coords(coords, input.shape[1], input.shape[0]);
flipcoords = [flip_coord(r, height) for r in sancoords]
narray = ocrolib.numpy2narray(input)
for rect in flipcoords:
ocrolib.iulib.fill_rect(narray, rect.x0, rect.y0, rect.x1, rect.y1, 255)
return ocrolib.narray2numpy(narray)
def process(self, input):
"""Blockout an image, using PIL. If any of
the parameters are -1 or less, use the
outer dimensions."""
height = input.shape[0]
pstr = self._params.get("boxes", "")
coords = get_coords(pstr)
if len(coords) == 0:
return input
sancoords = sanitise_coords(coords, input.shape[1], input.shape[0])
flipcoords = [flip_coord(r, height) for r in sancoords]
narray = ocrolib.numpy2narray(input)
for rect in flipcoords:
ocrolib.iulib.fill_rect(narray, rect.x0, rect.y0, rect.x1, rect.y1,
255)
return ocrolib.narray2numpy(narray)
def process(self, binary, boxes):
"""Recognize page text.
input: tuple of binary, input boxes
return: page data
"""
self.prepare()
pageheight, pagewidth = binary.shape
iulibbin = ocrolib.numpy2narray(binary)
out = [] # list of hocr strings
numcols = len(boxes.get("columns", []))
for i in range(numcols):
set_progress(self.logger, self.progress_func, i, numcols)
coords = boxes.get("columns")[i]
iulibcoords = (coords[0], pageheight - coords[3], coords[2],
pageheight - coords[1])
colimage = ocrolib.iulib.bytearray()
ocrolib.iulib.extract_subimage(colimage, iulibbin, *iulibcoords)
out.append(self.get_transcript(ocrolib.narray2numpy(colimage)))
set_progress(self.logger, self.progress_func, numcols, numcols)
self.cleanup()
return utils.merge_hocr(out)
def process(self, binary, boxes):
"""Recognize page text.
input: tuple of binary, input boxes
return: page data
"""
self.prepare()
pageheight, pagewidth = binary.shape
iulibbin = ocrolib.numpy2narray(binary)
out = [] # list of hocr strings
numcols = len(boxes.get("columns", []))
for i in range(numcols):
set_progress(self.logger, self.progress_func, i, numcols)
coords = boxes.get("columns")[i]
iulibcoords = (
coords[0], pageheight - coords[3], coords[2],
pageheight - coords[1])
colimage = ocrolib.iulib.bytearray()
ocrolib.iulib.extract_subimage(colimage, iulibbin, *iulibcoords)
out.append(self.get_transcript(ocrolib.narray2numpy(colimage)))
set_progress(self.logger, self.progress_func, numcols, numcols)
self.cleanup()
return utils.merge_hocr(out)
def segment_portion(self, portion, dx, dy, pheight):
"""Segment a single-column chunk."""
page_seg = self._segmenter.segment(ocrolib.narray2numpy(portion))
return self.extract_boxes(self._regions, page_seg, dx, dy, pheight)
def segment_portion(self, portion, dx, dy, pheight):
"""Segment a single-column chunk."""
page_seg = self._segmenter.segment(ocrolib.narray2numpy(portion))
return self.extract_boxes(self._regions, page_seg, dx, dy, pheight)