def index(request):
#from tesserwrap import Tesseract
#from PIL import Image
img = Image.open("/home/df/projects/django/nuspyp/tesseracttest/test.png")
tr = Tesseract()
tr.ocr_image(img)
img2 = dog(
filename='/home/df/projects/django/nuspyp/tesseracttest/source.pdf')
single_image = img2.sequence[0]
tr.ocr_image(single_image)
return HttpResponse(tr.get_text())
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
try:
img = Image.open(StringIO(data))
except DecompressionBombWarning as dce:
log.debug("Image too large: %", dce)
return None
except IOError as ioe:
log.info("Unknown image format: %r", ioe)
return None
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
extractor.clear()
log.debug('OCR done: %s, %s characters extracted',
languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
if TESSDATA_PREFIX is None:
raise ValueError("Env TESSDATA_PREFIX is not set, OCR will not work.")
key, text = get_cache(data)
if text is not None:
return text
try:
img = Image.open(StringIO(data))
except Exception as ex:
log.debug("Failed to parse image internally: %r", ex)
return ""
# TODO: play with contrast and sharpening the images.
try:
languages = _get_languages(languages)
extractor = Tesseract(TESSDATA_PREFIX, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
log.debug("OCR done: %s, %s characters extracted", languages, len(text))
set_cache(key, text)
return text
except Exception as ex:
log.exception(ex)
return ""
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
if TESSDATA_PREFIX is None:
raise ValueError('Env TESSDATA_PREFIX is not set, OCR will not work.')
key, text = get_cache(data)
if text is not None:
return text
try:
img = Image.open(StringIO(data))
except Exception as ex:
log.debug('Failed to parse image internally: %r', ex)
return ''
# TODO: play with contrast and sharpening the images.
try:
languages = _get_languages(languages)
extractor = Tesseract(TESSDATA_PREFIX, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
log.debug('OCR done: %s, %s characters extracted', languages,
len(text))
set_cache(key, text)
return text
except Exception as ex:
log.exception(ex)
return ''
def parse_img():
im = Image.open("./temp.jpg") # the second one
im = im.filter(ImageFilter.MedianFilter())
enhancer = ImageEnhance.Contrast(im)
im = enhancer.enhance(2)
im = im.convert('1')
im.save('./temp2.jpg')
tr = Tesseract(os.environ["TESSDATA_PREFIX"],"eng")
text = tr.ocr_image(Image.open('./temp2.jpg'))
return redirect('http://mailsnail.tech/api/notify')
def tesseract():
global semaphore
while(True):
if not os.path.exists("./output.png"):
break
semaphore.acquire()
img = Image.open("output.png")
tr = Tesseract("/usr/local/share")
text = tr.ocr_image(img)
print text
'''
subprocess.call(["tesseract", "output.png","out"])
'''
semaphore.release()
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
img = Image.open(StringIO(data))
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
log.debug('OCR done: %s, %s characters extracted', languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
img = Image.open(StringIO(data))
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
log.debug('OCR done: %s, %s characters extracted',
languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def pages(self):
for page in range(self.file.numPages):
img = WandImage(filename=self.path + ('[%s]' % page),
resolution=self.config['wand_resolution'])
img.compression_quality = self.config['wand_compression_quality']
temp = NamedTemporaryFile(suffix='.jpg')
# Passing temp as file kwargs does not work for some reason.
# So we just pass the filename.
img.save(filename=temp.name)
# Reopen the image file as PIL object
img = Image.open(temp.name)
# Run tesseract
tr = Tesseract()
result = tr.ocr_image(img)
temp.close()
yield result
def ocr(info):
cv2.imwrite('../fig/info.jpg', info)
img = Image.open('../fig/info.jpg')
tr = Tesseract(datadir='../data', lang='eng')
text = tr.ocr_image(img)
print(text)
def handleFrameForTaskB(self, frame, regionCoordinates):
try:
coordinates = list()
for point in regionCoordinates:
coordinates.append(
[point[0] * frame.shape[1], point[1] * frame.shape[0]])
coordinates = np.int0(coordinates)
frame = cv2.drawContours(frame, [coordinates], 0, (0, 255, 0), 2)
warped = four_point_transform(frame, coordinates)
shrunk = cv2.cvtColor(warped[:, int(warped.shape[1] / 10):],
cv2.COLOR_BGR2GRAY)
scale = 6
shrunk = cv2.resize(
shrunk, (shrunk.shape[1] * scale, shrunk.shape[0] * scale),
interpolation=cv2.INTER_CUBIC)
_, shrunk = cv2.threshold(shrunk, 100, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)
shrunk = 255 - cv2.dilate(
255 - shrunk, np.ones((1, 1)), iterations=1)
num, features = cv2.connectedComponents(255 - shrunk)
plate = str()
corners = list()
for i in range(0, num):
area = np.sum((features == i))
if area > scale**2 * 2 * 25 and area < scale * 4 * 500:
rows = np.any(features == i, axis=1)
cols = np.any(features == i, axis=0)
rmin, rmax = np.where(rows)[0][[0, -1]]
cmin, cmax = np.where(cols)[0][[0, -1]]
corners.append([rmin, cmin, rmax, cmax])
corners = np.array(corners)
idx = np.argsort(corners[:, 1])
sorted_corners = corners[idx]
for corner in sorted_corners:
minx = corner[0] - 2
miny = corner[1] - 2
maxx = corner[2] + 2
maxy = corner[3] + 2
if minx < 0:
minx = 0
if miny < 0:
miny = 0
snip = features[minx:maxx, miny:maxy]
if snip.shape[1] > snip.shape[0]:
continue
snip = cv2.erode(snip.astype(np.uint8),
np.ones((5, 5)),
iterations=1)
im = Image.fromarray(np.uint8(snip))
tr = Tesseract(datadir="/usr/share/tessdata")
letter = tr.ocr_image(im).rstrip()
for l in letter:
if l.isalnum():
letter = l
plate += letter.capitalize()
alphs = "".join(itertools.takewhile(str.isalpha, plate))
nums = plate[len(alphs):]
if len(alphs) == 2:
plate = alphs[0] + "-" + alphs[1] + "-" + nums
elif len(alphs) == 5:
plate = alphs[:3] + "-" + alphs[3:] + "-" + nums
else:
diffs = list()
alphscorners = sorted_corners[:len(alphs)]
for i in range(len(alphscorners)):
if sorted_corners[i][1] == alphscorners[-1][1]:
break
diffs.append(sorted_corners[i + 1][1] -
sorted_corners[i][3])
cuts = np.array(diffs) > np.mean(diffs)
rev_cuts = cuts[::-1]
for i in range(len(cuts[::-1])):
if (rev_cuts[i] == 1):
alphs = alphs[:len(cuts) - i] + "-" + alphs[len(cuts) -
i:]
plate = alphs + "-" + nums
if len(plate) < 5:
return None
elif len(plate) > 11:
return None
elif plate.count("-") > 2:
return None
elif plate.count("-") < 2:
return None
else:
return plate
except Exception as exception:
return None
warp = warp[::-1]
#print(warp)
fig = plt.figure(1, (10,10))
grid = ImageGrid(fig, 111,
nrows_ncols = (17, 10),
axes_pad=0.1,
aspect=True,
)
tsOutput = []
for i in range(0,162):
#print warp[i]
image = warp[i]
(h, w) = image.shape[:2]
center = (w / 2, h / 2)
M = cv2.getRotationMatrix2D(center, 270, 1.0)
rotated = cv2.warpAffine(image, M, (w, h))
letter=cv2.flip(rotated,1)
grid[i].imshow(letter)
im = Image.fromarray(letter)
text = tr.ocr_image(im).replace("\n","")
if text == "":
text = " "
tsOutput.append(text)
#print tr.ocr_image(im),
#print(tr.ocr_image(im),end = "")
print "".join(tsOutput)
plt.show()
cv2.destroyAllWindows()
def ocr(info):
cv2.imwrite('../fig/info.jpg', info)
img = Image.open('../fig/info.jpg')
tr = Tesseract(datadir='../data', lang='eng')
text = tr.ocr_image(img)
print(text)
#print height, " ", width
#sleep(5)
#gray = cv2.resize(image, dim, interpolation = cv2.INTER_AREA)
#height, width = gray.shape[:2]
# Our operations on the frame come here, doing some manipulation to reduce effects from surrounding objects,
#light&shadow play, etc
gray = cv2.medianBlur(gray, 5)
#gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
gray = cv2.bilateralFilter(gray, 5, 25, 25)
#cv2.rectangle(gray, (width/2-150, height/2-150), (width/2+200, height/2+150), (255,0,0), 2)
cv2.imshow('frame',gray) #showing image
#gray = gray.copy()[height/2-150:height/2+150,width/2-150:width/2+200]
image = Image.fromarray(gray) #turning array into image
speech = ocr.ocr_image(image) #ocr-ing image by Tesseract
#print type(speech)
#speech = ocr.get_utf8_text()
if ocr.get_mean_confidence() >= 63 and len(speech.strip()) >1:
tts.say(str(speech))
print speech
sleep(1)
#robot speech
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# When everything done, release the capture
#cap.release()
camProxy.unsubscribe(videoClient)
cv2.destroyAllWindows()
#print(warp)
fig = plt.figure(1, (10, 10))
grid = ImageGrid(
fig,
111,
nrows_ncols=(17, 10),
axes_pad=0.1,
aspect=True,
)
tsOutput = []
for i in range(0, 162):
#print warp[i]
image = warp[i]
(h, w) = image.shape[:2]
center = (w / 2, h / 2)
M = cv2.getRotationMatrix2D(center, 270, 1.0)
rotated = cv2.warpAffine(image, M, (w, h))
letter = cv2.flip(rotated, 1)
grid[i].imshow(letter)
im = Image.fromarray(letter)
text = tr.ocr_image(im).replace("\n", "")
if text == "":
text = " "
tsOutput.append(text)
#print tr.ocr_image(im),
#print(tr.ocr_image(im),end = "")
print "".join(tsOutput)
plt.show()
cv2.destroyAllWindows()
from tesserwrap import Tesseract
from PIL import Image
tr = Tesseract("/usr/local/share") # this is slow
im = Image.open("test2.png")
text = tr.ocr_image(im)
print text
words = text.split()
for thing in words:
if thing == "Arlington":
print "found ittt"
