def test_image_to_data__pandas_support(test_file_small):
with pytest.raises(TSVNotSupported):
image_to_data(test_file_small, output_type=Output.DATAFRAME)
import cv2
import pytesseract
from pytesseract import Output
img = cv2.imread('page_1.jpg')
d = pytesseract.image_to_data(img, output_type=Output.DICT)
print(d.keys())
n_boxes = len(d['text'])
for i in range(n_boxes):
if int(d['conf'][i]) > 60:
(x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i],
d['height'][i])
img = cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv2.imwrite("C:/New folder/New Vision Soft/OCR/ITR/box_img.jpg", img)
cv2.imshow('img', img)
cv2.waitKey(0)
def camIdMotor():
font = cv2.FONT_HERSHEY_SIMPLEX
def image_smoothening(img):
ret1, th1 = cv2.threshold(img, BINARY_THREHOLD, 255, cv2.THRESH_BINARY)
ret2, th2 = cv2.threshold(th1, 0, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)
blur = cv2.GaussianBlur(th2, (1, 1), 0)
ret3, th3 = cv2.threshold(blur, 0, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)
return th3
def remove_noise_and_smooth(img):
print('Removing noise and smoothening image')
# img = cv2.imread(file_name, 0)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
filtered = cv2.adaptiveThreshold(gray.astype(np.uint8), 255,
cv2.ADAPTIVE_THRESH_MEAN_C,
cv2.THRESH_BINARY, 41, 3)
kernel = np.ones((1, 1), np.uint8)
opening = cv2.morphologyEx(filtered, cv2.MORPH_OPEN, kernel)
closing = cv2.morphologyEx(opening, cv2.MORPH_CLOSE, kernel)
img = image_smoothening(gray)
or_image = cv2.bitwise_or(gray, closing)
return or_image
def gaussian_kernel(size, sigma=1):
size = int(size) // 2
x, y = np.mgrid[-size:size + 1, -size:size + 1]
normal = 1 / (2.0 * np.pi * sigma**2)
g = np.exp(-((x**2 + y**2) / (2.0 * sigma**2))) * normal
return g
def sobel_filters(img):
Kx = np.array([[-1, 0, 1], [-2, 0, 2], [-1, 0, 1]], np.float32)
Ky = np.array([[1, 2, 1], [0, 0, 0], [-1, -2, -1]], np.float32)
Ix = ndimage.filters.convolve(img, Kx)
Iy = ndimage.filters.convolve(img, Ky)
G = np.hypot(Ix, Iy)
G = G / G.max() * 255
theta = np.arctan2(Iy, Ix)
return (G, theta)
def non_max_suppression(img, D):
M, N = img.shape
Z = np.zeros((M, N), dtype=np.int32)
angle = D * 180. / np.pi
angle[angle < 0] += 180
for i in range(1, M - 1):
for j in range(1, N - 1):
try:
q = 255
r = 255
# angle 0
if (0 <= angle[i, j] < 22.5) or (157.5 <= angle[i, j] <=
180):
q = img[i, j + 1]
r = img[i, j - 1]
# angle 45
elif (22.5 <= angle[i, j] < 67.5):
q = img[i + 1, j - 1]
r = img[i - 1, j + 1]
# angle 90
elif (67.5 <= angle[i, j] < 112.5):
q = img[i + 1, j]
r = img[i - 1, j]
# angle 135
elif (112.5 <= angle[i, j] < 157.5):
q = img[i - 1, j - 1]
r = img[i + 1, j + 1]
if (img[i, j] >= q) and (img[i, j] >= r):
Z[i, j] = img[i, j]
else:
Z[i, j] = 0
except IndexError as e:
pass
return Z
def threshold(img):
highThreshold = img.max() * highthreshold
lowThreshold = highThreshold * lowthreshold
M, N = img.shape
res = np.zeros((M, N), dtype=np.int32)
weak = np.int32(weak_pixel)
strong = np.int32(strong_pixel)
strong_i, strong_j = np.where(img >= highThreshold)
zeros_i, zeros_j = np.where(img < lowThreshold)
weak_i, weak_j = np.where((img <= highThreshold)
& (img >= lowThreshold))
res[strong_i, strong_j] = strong
res[weak_i, weak_j] = weak
return (res)
def hysteresis(self, img):
M, N = img.shape
weak = self.weak_pixel
strong = self.strong_pixel
for i in range(1, M - 1):
for j in range(1, N - 1):
if (img[i, j] == weak):
try:
if ((img[i + 1, j - 1] == strong)
or (img[i + 1, j] == strong)
or (img[i + 1, j + 1] == strong)
or (img[i, j - 1] == strong)
or (img[i, j + 1] == strong)
or (img[i - 1, j - 1] == strong)
or (img[i - 1, j] == strong)
or (img[i - 1, j + 1] == strong)):
img[i, j] = strong
else:
img[i, j] = 0
except IndexError as e:
pass
return img
def hysteresis(img):
M, N = img.shape
weak = weak_pixel
strong = strong_pixel
for i in range(1, M - 1):
for j in range(1, N - 1):
if (img[i, j] == weak):
try:
if ((img[i + 1, j - 1] == strong)
or (img[i + 1, j] == strong)
or (img[i + 1, j + 1] == strong)
or (img[i, j - 1] == strong)
or (img[i, j + 1] == strong)
or (img[i - 1, j - 1] == strong)
or (img[i - 1, j] == strong)
or (img[i - 1, j + 1] == strong)):
img[i, j] = strong
else:
img[i, j] = 0
except IndexError as e:
pass
return img
def detectarPlaca(img):
##TESTING
# blurred_image = gaussian_blur(image, kernel_size=9, verbose=False)
# img_smoothed = convolve(img, gaussian_kernel(kernel_size, sigma))
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# gradientMat, thetaMat = sobel_filters(img_smoothed)
# nonMaxImg = non_max_suppression(gradientMat, thetaMat)
# thresholdImg = threshold(nonMaxImg)
# img_final = hysteresis(thresholdImg)
# imgNoise = remove_noise_and_smooth(img)
# grayNew= unsharp_mask(img)
# im_resized(img, dpi=(300, 300)) # best for OCR
# img = cv2.resize(img, (300, 300))
# img = cv2.medianBlur(img, 5)
# gray = cv2.blur(gray, (5, 5))
# gray= cv2.medianBlur(gray, 5)
# gray= img.gaussian_kernel(5)
gray = cv2.GaussianBlur(gray, (5, 5), 0)
# gray = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]
# ret, thresh1 = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY)
# ret, thresh2 = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY_INV)
# ret, thresh3 = cv2.threshold(gray, 127, 255, cv2.THRESH_TRUNC)
# ret, thresh4 = cv2.threshold(gray, 127, 255, cv2.THRESH_TOZERO)
# ret, thresh5 = cv2.threshold(gray, 127, 255, cv2.THRESH_TOZERO_INV)
## OTHER
# th2 = cv2.adaptiveThreshold(img, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 11, 2)
# th3 = cv2.adaptiveThreshold(img, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 11, 2)
# ret3, th3 = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
canny = cv2.Canny(gray, 50, 150)
# kernel = np.ones((5, 5), np.uint8)
canny = cv2.dilate(canny, None, iterations=2)
# canny = cv2.erode(canny, None, iterations=1)
kernel = np.ones((5, 5), np.uint8)
# canny = cv2.morphologyEx(canny, cv2.MORPH_CLOSE, None)
# canny = cv2.morphologyEx(canny, cv2.MORPH_GRADIENT,None)
# canny = cv2.dilate(canny, None, iterations=2)
# canny= cv2.bitwise_not(canny)
# canny = cv2.erode(canny, kernel, iterations=1)
return canny
def unsharp_mask(image,
kernel_size=(5, 5),
sigma=1.0,
amount=1.0,
threshold=0):
"""Return a sharpened version of the image, using an unsharp mask."""
blurred = cv2.GaussianBlur(image, kernel_size, sigma)
sharpened = float(amount + 1) * image - float(amount) * blurred
sharpened = np.maximum(sharpened, np.zeros(sharpened.shape))
sharpened = np.minimum(sharpened, 255 * np.ones(sharpened.shape))
sharpened = sharpened.round().astype(np.uint8)
if threshold > 0:
low_contrast_mask = np.absolute(image - blurred) < threshold
np.copyto(sharpened, image, where=low_contrast_mask)
return sharpened
def binarizacion(img):
# img = cv2.imread('gradient.png', 0)
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
ret, thresh1 = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)
ret, thresh2 = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY_INV)
ret, thresh3 = cv2.threshold(img, 127, 255, cv2.THRESH_TRUNC)
ret, thresh4 = cv2.threshold(img, 127, 255, cv2.THRESH_TOZERO)
ret, thresh5 = cv2.threshold(img, 127, 255, cv2.THRESH_TOZERO_INV)
# titles = ['Original Image', 'BINARY', 'BINARY_INV', 'TRUNC', 'TOZERO', 'TOZERO_INV']
# images = [img, thresh1, thresh2, thresh3, thresh4, thresh5]
return thresh5
def conTornos(src):
# gris = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)
# Aplicar suavizado Gaussiano
# gauss = cv2.GaussianBlur(gris, (5, 5), 0)
# canny = cv2.Canny(gauss, 50, 150)
# imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# ret, thresh = cv2.threshold(imgray, 127, 255, 0)
# im2, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
# gray = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)
# gray = cv2.GaussianBlur(gray, (7, 7), 3)
# t, dst = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_TRIANGLE)
# gray = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)
# _, th = cv2.threshold(gray, 100, 255, cv2.THRESH_BINARY)
# canny = cv2.Canny(th, 50, 150)
# img1, contornos1, hierarchy1 = cv2.findContours(th, cv2.RETR_EXTERNAL,
# cv2.CHAIN_APPROX_NONE)
# img2, contornos2, hierarchy2 = cv2.findContours(th, cv2.RETR_EXTERNAL,
# cv2.CHAIN_APPROX_SIMPLE)
# contornos1, hierarchy1 = cv2.findContours(th, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
gray = cv2.cvtColor(src, cv2.COLOR_BGR2GRAY)
# gray = cv2.blur(gray, (3, 3))
gray = cv2.bilateralFilter(gray, 11, 17, 17) # Blur to reduce noise
th3 = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, \
cv2.THRESH_BINARY, 11, 2)
canny = cv2.Canny(gray, 150, 200)
# canny = cv2.dilate(canny, None, iterations=1)
canny = cv2.morphologyEx(canny, cv2.MORPH_OPEN, None)
return canny
def maskNew2(img):
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # grayscale
_, thresh = cv2.threshold(gray, 150, 255,
cv2.THRESH_BINARY_INV) # threshold
kernel = cv2.getStructuringElement(cv2.MORPH_CROSS, (3, 3))
dilated = cv2.dilate(thresh, kernel, iterations=13) # dilate
return dilated
def maskNew1(img):
gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
#canny = cv2.Canny(gray_img, 100, 200)
# gray_img = canny
ret1, th1 = cv2.threshold(gray_img, 127, 255, cv2.THRESH_BINARY)
# th3 = cv2.adaptiveThreshold(gray_img, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 11, 2)
# global thresholding
ret1, th1 = cv2.threshold(gray_img, 127, 255, cv2.THRESH_BINARY)
# Otsu's thresholding
ret2, th2 = cv2.threshold(gray_img, 0, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)
# Otsu's thresholding after Gaussian filtering
blur = cv2.GaussianBlur(gray_img, (5, 5), 0)
ret3, th3 = cv2.threshold(blur, 0, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)
##other
ret, thresh1 = cv2.threshold(gray_img, 127, 255, cv2.THRESH_BINARY)
ret, thresh2 = cv2.threshold(gray_img, 127, 255, cv2.THRESH_BINARY_INV)
ret, thresh3 = cv2.threshold(gray_img, 127, 255, cv2.THRESH_TRUNC)
ret, thresh4 = cv2.threshold(gray_img, 127, 255, cv2.THRESH_TOZERO)
ret, thresh5 = cv2.threshold(gray_img, 127, 255, cv2.THRESH_TOZERO_INV)
return th2
def rescalingNew(imgNew):
img = cv2.resize(imgNew,
None,
fx=1.2,
fy=1.2,
interpolation=cv2.INTER_CUBIC)
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# kernel = np.ones((1, 1), np.uint8)
# img = cv2.dilate(img, kernel, iterations=1)
# img = cv2.erode(img, kernel, iterations=1)
# cv2.threshold(cv2.bilateralFilter(img, 5, 75, 75), 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]
# cv2.threshold(cv2.medianBlur(img, 3), 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]
# cv2.adaptiveThreshold(cv2.bilateralFilter(img, 9, 75, 75), 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
# cv2.THRESH_BINARY, 31, 2)
cv2.adaptiveThreshold(cv2.medianBlur(img, 3), 255,
cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY, 31, 2)
return img
def maskNew(img):
frame = cv2.resize(img, (400, 300), interpolation=cv2.INTER_CUBIC)
hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
lowerBlue = np.array([100, 50, 50])
upperBlue = np.array(
[130, 255, 255]
) # cv2.adaptiveThreshold(cv2.medianBlur(img, 3), 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 31, 2)
mask = cv2.inRange(hsv, lowerBlue, upperBlue)
return mask
def detectaCode(img):
gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray_img = cv2.bitwise_not(gray_img)
return gray_img
def umbralChangeBack(img):
gray_img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
_, thresh = cv2.threshold(gray_img, 127, 255,
cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
img_contours = cv2.findContours(thresh, cv2.RETR_TREE,
cv2.CHAIN_APPROX_SIMPLE)[-2]
img_contours = sorted(img_contours, key=cv2.contourArea)
for i in img_contours:
if cv2.contourArea(i) > 100:
break
mask = np.zeros(img.shape[:2], np.uint8)
cv2.drawContours(mask, [i], -1, 255, -1)
new_img = cv2.bitwise_and(img, img, mask=mask)
return new_img
def contrast(img):
return cv2.addWeighted(img, 1.5, np.zeros(img.shape, img.dtype), 0, 0)
def focusing(val):
value = (val << 4) & 0x3ff0
data1 = (value >> 8) & 0x3f
data2 = value & 0xf0
os.system("i2cset -y 0 0x0c %d %d" % (data1, data2))
def sobel(img):
img_gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
img_sobel = cv2.Sobel(img_gray, cv2.CV_16U, 1, 1)
return cv2.mean(img_sobel)[0]
def laplacian(img):
img_gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
img_sobel = cv2.Laplacian(img_gray, cv2.CV_16U)
return cv2.mean(img_sobel)[0]
# get grayscale image
def get_grayscale(image):
img = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
ret, thresh2 = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY_INV)
# Aplicar suavizado Gaussiano
gauss = cv2.GaussianBlur(thresh2, (5, 5), 0)
# canny = cv2.Canny(gauss, 100, 200)
# canny= auto_canny(img)
return gauss
def get_grayscaleNEW(image):
return cv2.bilateralFilter(image, 11, 17, 17)
# noise removal
def remove_noise(image):
return cv2.medianBlur(image, 5)
# thresholding
def thresholding(image):
return cv2.threshold(image, 0, 255,
cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]
def thresHolding(img):
return cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)
# dilation
def dilate(image):
kernel = np.ones((5, 5), np.uint8)
return cv2.dilate(image, kernel, iterations=1)
# erosion
def erode(image):
kernel = np.ones((5, 5), np.uint8)
return cv2.erode(image, kernel, iterations=1)
# opening - erosion followed by dilation
def opening(image):
kernel = np.ones((5, 5), np.uint8)
return cv2.morphologyEx(image, cv2.MORPH_OPEN, kernel)
# canny edge detection
def canny(image):
return cv2.Canny(image, 100, 200)
# skew correction
def auto_canny(image, sigma=0.33):
# compute the median of the single channel pixel intensities
v = np.median(image)
# apply automatic Canny edge detection using the computed median
lower = int(max(0, (1.0 - sigma) * v))
upper = int(min(255, (1.0 + sigma) * v))
edged = cv2.Canny(image, lower, upper)
# return the edged image
return edged
def deskew(image):
coords = np.column_stack(np.where(image > 0))
angle = cv2.minAreaRect(coords)[-1]
if angle < -45:
angle = -(90 + angle)
else:
angle = -angle
(h, w) = image.shape[:2]
center = (w // 2, h // 2)
M = cv2.getRotationMatrix2D(center, angle, 1.0)
rotated = cv2.warpAffine(image,
M, (w, h),
flags=cv2.INTER_CUBIC,
borderMode=cv2.BORDER_REPLICATE)
return rotated
# template matching
def match_template(image, template):
return cv2.matchTemplate(image, template, cv2.TM_CCOEFF_NORMED)
ap = argparse.ArgumentParser()
ap.add_argument("-v", "--video", help="path to the (optional) video file")
args = vars(ap.parse_args())
# if the video path was not supplied, grab the reference to the
# camera
if not args.get("video", False):
vs = VideoStream(src=0).start()
# vs = cv2.VideoCapture(0)
time.sleep(2.0)
# otherwise, load the video
else:
vs = cv2.VideoCapture(args["video"])
# keep looping over the frames
while True:
# check to see if we have reached the end of the
# video
# time.sleep(2.0)
frame = vs.read()
frame = frame[1] if args.get("video", False) else frame
if frame is None:
break
# length = int(vs.get(cv2.cv.CV_CAP_PROP_FRAME_COUNT))
# width = int(vs.get(cv2.cv.CV_CAP_PROP_FRAME_WIDTH))
# height = int(vs.get(cv2.cv.CV_CAP_PROP_FRAME_HEIGHT))
# fps = vs.get(cv2.cv.CV_CAP_PROP_FPS)
# grab the current frame and then handle if the frame is returned
# from either the 'VideoCapture' or 'VideoStream' object,
# respectively
# frame = thresholding(frame)
# frame = opening(frame)
# frame = canny(frame)
# frame = canny(frameOriginal)
# frameModificado = get_grayscale(frame)
# frameModificado = binarizacion(frame)
frameModificado = detectarPlaca(frame)
# frame= image_smoothening(frame)
# frame= auto_canny(frame)
# frame = sobel(frame)
# frame = laplacian(frame)
# frame = umbralChangeBack(frame)
# frame = detectaCode(frame)
# frame = get_grayscale(frame)
# frame = thresHolding(frame)
# frame = maskNew1(frame)
# frame = contrast(frame)
# frame = conTornos(frame)
# frame = maskNew1(frame)
# frame= rescalingNew(frame)
# contornos1,hierarchy1 = cv2.findContours(frameOriginal, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
cnts, _ = cv2.findContours(frameModificado, cv2.RETR_LIST,
cv2.CHAIN_APPROX_SIMPLE) # OpenCV 4
# print('Número de contornos encontrados: ', len(cnts))
for c in cnts:
area = cv2.contourArea(c)
if area > 1500 and area < 6000:
# print(area)
# nuevoContorno = cv2.convexHull(c)
perimeter = cv2.arcLength(c, True)
epsilon = 0.1 * cv2.arcLength(c, True)
approx = cv2.approxPolyDP(c, epsilon, True)
hull = cv2.convexHull(c)
M = cv2.moments(c)
if (M["m00"] == 0): M["m00"] = 1
x = int(M["m10"] / M["m00"])
y = int(M['m01'] / M['m00'])
cv2.drawContours(frame, [approx], 0, (255, 0, 0), 3)
# cv2.drawContours(frame, c, -1, (0, 0, 255), 3)
# Dibujar el centro
cv2.circle(frame, (x, y), 3, (0, 0, 255), -1)
# cv2.drawContours(frame, cnts, -1, (0, 255, 0), 2)
# epsilon = 0.01 * cv2.arcLength(c, True)
# approx = cv2.approxPolyDP(c, epsilon, True)
# print(len(approx))
# x, y, w, h = cv2.boundingRect(approx)
# print('AREA', x, y, w, h)
# placa = frameModificado[y:y + h, x:x + w]
# textNew = pytesseract.image_to_string(placa, config='--psm 11')
# print('PLACA: ', textNew)
# textCycle = pytesseract.image_to_string([nuevoContorno], config='--psm 11')
# print(textCycle)
custom_oem_psm_config = r'--oem 3 --psm 6'
# config = '--psm 11'
d = pytesseract.image_to_data(frameModificado,
output_type=Output.DICT,
config=custom_oem_psm_config)
#print(d.keys())
print(d['text'])
## EXAMPLE
text = pytesseract.image_to_string(frameModificado,
config=custom_oem_psm_config)
# print("Detected Number is:", text)
x = 0
y = 0
n_boxes = len(d['text'])
for i in range(n_boxes):
if int(d['conf'][i]) > 60:
(x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i],
d['height'][i])
frameModificado = cv2.rectangle(frameModificado, (x, y),
(x + w, y + h), (0, 255, 0), 2)
cv2.putText(frameModificado, '{}'.format(d['text']), (x - 60, y + 160),
font, 1.00, (0, 0, 0), 2, cv2.LINE_AA)
cv2.imshow('Frame', frame)
cv2.imshow('Frame_Modificado', frameModificado)
# cv2.imshow("Frame_Original", frameOriginal)
# Salir con 'ESC'
patronNew = re.compile('^FKI[0-9][0-9][0-9][0-9][0-9]')
tempLength = d['text']
print(len(d['text']))
for t in d['text']:
findText = re.search(patronNew, t, flags=0)
# print("Iteracion: " + t)
if findText:
print("Detected Number is:", findText.group(0))
break
# findText = re.search(patronNew, text, flags=0)
# print(findText.group(0))
if findText:
## print("Detected Number is:", findText.group(0))
break
k = cv2.waitKey(5) & 0xFF
if k == 27:
break
# if we are not using a video file, stop the video file stream
if not args.get("video", False):
vs.stop()
# otherwise, release the camera pointer
else:
vs.release()
# close all windows
cv2.destroyAllWindows()
# waitTime = findText.group(0)
# print('waitTime: ' + waitTime)
return findText.group(0)
titles = [
'Original Image', 'BINARY', 'BINARY_INV', 'TRUNC', 'TOZERO', 'TOZERO_INV'
]
images = [img, thresh1, thresh2, thresh3, thresh4, thresh5]
for i in range(6):
plt.subplot(2, 3, i + 1)
plt.imshow(images[i], 'gray', vmin=0, vmax=255)
plt.title(titles[i])
plt.xticks([]), plt.yticks([])
plt.show()
custom_config = r'--oem 3 --psm 6 -l kor+eng'
words = pytesseract.image_to_data(thresh4,
config=custom_config,
output_type=Output.DICT) ## pytesseract
n_boxes = len(words['text'])
print(n_boxes)
for i in range(n_boxes):
if int(words['conf'][i]) > 60:
(x, y, w, h) = (words['left'][i], words['top'][i], words['width'][i],
words['height'][i])
cv.rectangle(img, (x, y), (x + w, y + h), (0, 0, 255), 1)
cv.putText(img, words['text'][i], (x, y), cv.FONT_HERSHEY_COMPLEX, 0.5,
(0, 0, 0), 1)
img_crop = img[y:y + h, x:x + w]
cv.imshow('img_crop', img_crop)
cv.imshow('Resource', img)
cv.waitKey(0)
def __init__(self,
dir_to_find=None,
oem=None,
psm=None,
parent=None,
filter_image_selected=None):
from os import scandir, path
# App Message
if parent == None:
self.parent = self
self.parent.info = self.info
#check exist and set path of EXE Tesseract
exe_tesseract = r'C:\Program Files\Tesseract-OCR\tesseract.exe'
if not path.exists(exe_tesseract):
self.parent.info('File not exist: ' + exe_tesseract)
return
pytesseract.pytesseract.tesseract_cmd = exe_tesseract
# Adding custom options
if not oem == None and not psm == None:
custom_config = r'--oem ' + str(oem) + ' --psm ' + str(psm[:2])
elif not oem == None:
custom_config = r'--oem ' + str(oem)
custom_config = r' --psm' + str(psm)
else:
custom_config = r'--oem 3 '
# Select Photo Directory
if dir_to_find == None:
self.select_directory()
else:
self.dir_to_find = dir_to_find
result = {}
correct = 0
incorrect = 0
# Create TXT File with Results
file_txt = open(path.join(self.dir_to_find, 'Result_OCR_Scan.txt'),
'w')
file_txt.write('Configuration ' + custom_config + chr(10))
file_txt.write('Filter ' + filter_image_selected + chr(10) + chr(10))
for file in scandir(self.dir_to_find):
# extract name file and extension separate
name, ext = path.splitext(file.name)
# Find coordinates from a picture file
if ext.upper() == '.JPG' or ext.upper() == '.JPEG' or ext.upper(
) == '.PNG':
find = False
incorrect += 1
day = ' No Encontrado'
hour = ''
lat = ''
lon = ''
# Filter Image for a good reading
self.filter_image(file.path, filter_image_selected)
read = pytesseract.image_to_data(self.img,
output_type=Output.DICT)
for i in range(len(read['text'])):
if read['text'][i] == 'GARMIN':
find = True
day = read['text'][i + 1]
hour = read['text'][i + 2]
lat = float(read['text'][i + 3])
lon = float(read['text'][i + 4])
self.geotag_photo(file.path, lat, lon)
incorrect -= 1
correct += 1
break
result[file.path] = {
'day': day,
'hour': hour,
'lat': lat,
'lon': lon,
'find': find
}
text_result = file.path + " " + day + " " + hour + " " + str(
lat) + " " + str(lon)
file_txt.write(text_result + chr(10))
self.parent.info(text_result)
#self.parent.info(result)
self.parent.info('GeoLocalizados : ' + str(correct))
self.parent.info('No GeoLocalizados: ' + str(incorrect))
file_txt.write('GeoLocalizados : ' + str(correct) + chr(10))
file_txt.write('No GeoLocalizados: ' + str(incorrect) + chr(10))
file_txt.close()
def imgtodata(i):
j = pytesseract.image_to_data(i, output_type=Output.DICT)
#print(j.keys())
k = len(j['text'])
return j, k
"""
Created on Tue Mar 3 16:32:52 2020
@author: Namita
Simple program to detect text and boxing it using pytessersct
Usage : python box.py
"""
import cv2
import pytesseract
from pytesseract import Output
img_cv = cv2.imread('image.jpg') # read image using image path
img = cv2.cvtColor(
img_cv, cv2.COLOR_BGR2RGB) # Because pytesseract works good on grayscale
d = pytesseract.image_to_data(
img, output_type=Output.DICT) #convert image to data object
# print(pytesseract.image_to_boxes(img))
# print(d['text'])
n_boxes = len(d['text'])
for i in range(n_boxes):
if int(d['conf'][i]) > 60:
(x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i],
d['height'][i])
img = cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv2.imshow('IMG.png', img) #display image
cv2.waitKey(
0
) #wait key zero means manual closing the window, You can add time in mils
from numpy import asarray
#ap = argparse.ArgumentParser()
#p.add_argument("-i" , "--testID.jpg" , help = "path to input image to be OCR'd", default = "stdout")
#ap.add_argument ("-o" , "--output" , help = "path to the output CSV file")
#ap.add_argument("-c" , "--min conf" , help = "minimum confidence value to filter weak text detection")
#ap.add_argument("-d" , "--dist-thresh" , type = float , default = 25.0)
#ap.add_argument("-s", "--min-size", type = int, default = 2 , help = "minimum cluster size")
#args = vars(ap.parse_args())
#np.random.seed(42)
pytesseract.pytesseract.tesseract_cmd = r'C:\Program Files\Tesseract-OCR\tesseract.exe'
custom_config = r' --oem 3 --psm 11'
images = cv2.imread(r"C:\Users\King alagbe\Downloads\testID.jpg")
images = cv2.cvtColor(images, cv2.COLOR_BGR2RGB)
height, width, t = images.shape
boxes = pytesseract.image_to_data(images , config= custom_config)
print(boxes)
for x, b in enumerate(boxes.splitlines()):
if x!=0:
#print(b)
b =b.split()
print(b)
if len(b) == 12:
x,y,w,h = int(b[6]), int(b[7]), int(b[8]), int(b[9])
cv2.rectangle(images, (x, y), (w + x, h + y),(225,0,0), 2)
cv2.putText(images, b[11], (x,y), cv2.FONT_HERSHEY_COMPLEX,0.3,(50,50,255), 1)
cv2.imshow("Result", images)
cv2.waitKey(0)
def bot():
# get window and rect
window = windows[index]
rect = windowRects[index]
#sepperate x,y,w,h of rect
x = rect[0]
y = rect[1]
w = rect[2]
h = rect[3]
#create bounding box: top and height are cut
bounding_box = (x, y + 100, w, h - 160)
waittime = 0
try:
while True:
if (keyboard.is_pressed("q")):
break
windowimg = ImageGrab.grab(bbox=bounding_box)
#conert to type L, means just black and white
windowimg = windowimg.convert('L')
#convert image to numpy array to work with pixels
img = np.array(windowimg)
# add Threshhold, just black and white
ret, img = cv2.threshold(img, 240, 255, cv2.THRESH_BINARY)
d = pytesseract.image_to_data(img,
output_type='data.frame')
cv2.imshow('screen', img)
if (cv2.waitKey(1) & 0xFF) == ord('q'):
cv2.destroyAllWindows()
break
# filter the results of tesseract for flyff
# these filter arent perfect yet
d = d[d['text'].apply(lambda x: isinstance(x, str))]
d = d[(d['conf'] != -1) & (d['conf'] > 2) &
(d['text'] != "") & (d['text'] != " ") &
(d['text'] != "NaN") &
(d['text'].apply(lambda x: len(x) > 2))]
# text in image is found
if (d.shape[0] > 0):
# get row of found text
row = d.iloc[0]
# + 20 because x + row[left] is the x-top left corner of text
monsterx = x + row['left'] + 20
# + 5 because y + row[top] is the y-top left corner of text
# + 100 because we did + 100 at bounding box earlier
monstery = y + row['top'] + 100 + 5
print("Found Monster in Window: " + str(window) +
" at x: " + str(monsterx) + " at y: " +
str(monstery))
#click
click(monsterx, monstery)
#spellcast
keyboard.press_and_release("c")
# general wait time for next analysis of image
time.sleep(0.1)
except KeyboardInterrupt:
exit()
def inventory_item_full_info(slot_i, slot_pos, slot_size, ss_path):
"""
Returns a tuple follow this format : (slot_i, item_name, amount) (int, str, int)
It uses the item_amount function
:param slot_i: <int> Slot index
:param slot_pos: <tuple> Format (X, Y)
:param slot_size: <int> Slot size (needed for item_amount screenshots)
:param ss_path: <str> Path to save screenshots (it spams them) Example : "./screenshots/"
"""
max_length = 400
slot_info = (slot_i, "null", "null"
) #Won't be changed if no items detected
pydirectinput.moveTo(slot_pos[0], slot_pos[1])
screenshot = pyautogui.screenshot(f"{ss_path}slot_{slot_i}.png",\
region=(slot_pos[0]+17,slot_pos[1]-30, max_length, 24))#Takes Screenshot
pixel = screenshot.getpixel(
(1, 0)) #First top left pixel indicates if we found an item
# item names are surrounded in a pink/black box, that we use to detect if one found
if (25 <= pixel[0] <= 45) and (0 <= pixel[1] <= 15) and (80 <= pixel[2] <=
100):
# More or less close to (35, 0, 90) that we're looking for
# If entered here, an item has been detected
width = max_length - 1
while width >= 0:
new_px = screenshot.getpixel((width, 0))
if (25 <= new_px[0] <= 45) and (0 <= new_px[1] <=
15) and (80 <= new_px[2] <= 100):
# Means we reached the box around item name, no extra space on the right anymore
break
width -= 10
print(width)
screenshot = pyautogui.screenshot(f"{ss_path}slot_{slot_i}.png",\
region=(slot_pos[0]+17,slot_pos[1]-30, width, 24))#New updated screenshot
# Better OCR as a result of cropping the image
content = pytesseract.image_to_data(screenshot)
# image_to_data separates all text found in words
item_name = ''
for x, bx in enumerate(
content.splitlines()): # Each line holds different data
if x != 0: # First line is a different format and holds no data we need
bx = bx.split()
if len(bx) == 12: # len is 12 when a word is found
item_name += f'{bx[11]} ' # word held at index 11 (last one)
pyautogui.moveTo(
64, 103) #Outside of inventory to not block item screenshot
amount = item_amount(slot_pos,
slot_size,
f'slot_{slot_i}_amount.png',
ss_path=ss_path)
# Calls item_amount to get the amount on this slot
slot_info = (slot_i, item_name, amount)
return slot_info
import mysql.connector
import sys
from gtts import gTTS
from PIL import Image
from pytesseract import image_to_string
from pytesseract import image_to_boxes
from pytesseract import image_to_osd
from pytesseract import image_to_data
from pytesseract import image_to_pdf_or_hocr
#from pytesseract import osd_to_dict
#from pytesseract import run_tesseract
#from pytesseract import run_and_get_output
img = Image.open("/home/san/aaa/example_03.png")
text = image_to_data(img)
print text
text = image_to_boxes(img)
print text
#======================================
# db= mysql.connector.connect(user="******",password="******",host="localhost",database="dsaa")
# mycursor = db.cursor()
# print """
# +=========================================+
# | Please select page no. of Book |
# +=========================================+
# """
def processImage():
readImage = cv2.imread('screenshot.png')
imageHeight, imageWidth = readImage.shape[:2]
lowerWidthBound = int(.29 * imageWidth) # 555px / 1920px
upperWidthBound = int(.375 * imageWidth) # 720px / 1920px
lowerHeightBound = int(.435 * imageHeight) # 615px / 1080px
upperHeightBound = int(.57 * imageHeight) # 470px / 1080px
# Crop and block out colors on the image
croppedImage = readImage[lowerHeightBound:upperHeightBound,
lowerWidthBound:upperWidthBound]
lowerColorBound = np.array([27, 27, 27], dtype="uint16")
upperColorBound = np.array([50, 50, 50], dtype="uint16")
colorMask = cv2.inRange(croppedImage, lowerColorBound, upperColorBound)
# Find text contours
imageContours, imageHierarchy = cv2.findContours(colorMask,
cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
# Create lists to hold our Tesseract data
possibleChars = []
possibleCharsData = []
# Process everything that may be text
for possibleText in imageContours:
# Create a bounding rectangle around the contour
[x, y, width, height] = cv2.boundingRect(possibleText)
# Draw rectangles around the contours on the original image
cv2.rectangle(colorMask, (x, y), (x + width, y + height),
(255, 0, 255), 0)
# Create our region of interest
regionOfInterest = colorMask[y:(y + height), x:(x + width)]
# Enlarge the region
enlargedImage = cv2.copyMakeBorder(regionOfInterest,
10,
10,
10,
10,
cv2.BORDER_CONSTANT,
value=[255, 255, 255])
# Run Tesseract OCT against the image to try and find the most likely keystroke
possibleChars.append(
pytesseract.image_to_string(
regionOfInterest,
lang='eng',
config=
'--psm 10 -c tessedit_char_whitelist=abcdefghijklmnopqrstuvwxyz'
))
possibleCharsData.append(
pytesseract.image_to_data(
regionOfInterest,
lang='eng',
config=
'--psm 10 -c tessedit_char_whitelist=abcdefghijklmnopqrstuvwxyz'
))
# Return all of the possible characters for the AFK key
return possibleChars, possibleCharsData
import pytesseract
from PIL import Image
import cv2
print(pytesseract.image_to_data(Image.open('./img/np1.jpeg')))
print(pytesseract.image_to_data(Image.open('./img/np2.jpeg')))
img = cv2.imread('./img/np1.jpeg')
print(pytesseract.image_to_string(img))
box = cv2.boxPoints(rect)
box = np.int0(box)
#check for 90 degrees rotation
if ((width / float(height)) > 1):
# crop a particular rectangle from the source image
cropped_image = cv2.getRectSubPix(carsample_mask, (width, height),
centre)
else:
# crop a particular rectangle from the source image
cropped_image = cv2.getRectSubPix(carsample_mask, (height, width),
centre)
# convert into grayscale
cropped_image = cv2.cvtColor(cropped_image, cv2.COLOR_BGR2GRAY)
# equalize the histogram
cropped_image = cv2.equalizeHist(cropped_image)
# resize to 260 cols and 63 rows. (Just something I have set as standard here)
cropped_image = cv2.resize(cropped_image, (260, 180))
cropped_images.append(cropped_image)
_ = plt.subplots_adjust(hspace=0.000)
number_of_subplots = len(cropped_images)
for i, v in enumerate(range(number_of_subplots)):
v = v + 1
#ax1 = plt.subplot(number_of_subplots,1,v)
showfig(cropped_images[i], plt.get_cmap('gray'))
plt.show()
for img in cropped_images:
data = pytesseract.image_to_data(Image.fromarray(img), output_type='dict')
print(data)
def ocr_img(img: np.array,
input_file: str,
search_str: str,
highlight_readable_text: bool = False,
action: str = 'Highlight',
show_comparison: bool = False,
generate_output: bool = True):
"""Scans an image buffer or an image file.
Pre-processes the image.
Calls the Tesseract engine with pre-defined parameters.
Calculates the confidence score of the image grabbed content.
Draws a green rectangle around readable text items having a confidence score > 30.
Searches for a specific text.
Highlight or redact found matches of the searched text.
Displays a window showing readable text fields or the highlighted or redacted text.
Generates the text content of the image.
Prints a summary to the console."""
# If image source file is inputted as a parameter
if input_file:
# Reading image using opencv
img = cv2.imread(input_file)
# Preserve a copy of this image for comparison purposes
initial_img = img.copy()
highlighted_img = img.copy()
# Convert image to binary
bin_img = convert_img2bin(img)
# Calling Tesseract
# Tesseract Configuration parameters
# oem --> OCR engine mode = 3 >> Legacy + LSTM mode only (LSTM neutral net mode works the best)
# psm --> page segmentation mode = 6 >> Assume as single uniform block of text (How a page of text can be analyzed)
config_param = r'--oem 3 --psm 6'
# Feeding image to tesseract
details = pytesseract.image_to_data(bin_img,
output_type=Output.DICT,
config=config_param,
lang='eng')
# The details dictionary contains the information of the input image
# such as detected text, region, position, information, height, width, confidence score.
ss_confidence = calculate_ss_confidence(details)
boxed_img = None
# Total readable items
ss_readable_items = 0
# Total matches found
ss_matches = 0
for seq in range(len(details['text'])):
# Consider only text fields with confidence score > 30 (text is readable)
if float(details['conf'][seq]) > 30.0:
ss_readable_items += 1
# Draws a green rectangle around readable text items having a confidence score > 30
if highlight_readable_text:
(x, y, w, h) = (details['left'][seq], details['top'][seq],
details['width'][seq], details['height'][seq])
boxed_img = cv2.rectangle(img, (x, y), (x + w, y + h),
(0, 255, 0), 2)
# Searches for the string
if search_str:
results = re.findall(search_str, details['text'][seq],
re.IGNORECASE)
for result in results:
ss_matches += 1
if action:
# Draw a red rectangle around the searchable text
(x, y, w,
h) = (details['left'][seq], details['top'][seq],
details['width'][seq], details['height'][seq])
# Details of the rectangle
# Starting coordinate representing the top left corner of the rectangle
start_point = (x, y)
# Ending coordinate representing the botton right corner of the rectangle
end_point = (x + w, y + h)
# Color in BGR -- Blue, Green, Red
if action == "Highlight":
color = (0, 255, 255) # Yellow
elif action == "Redact":
color = (0, 0, 0) # Black
# Thickness in px (-1 will fill the entire shape)
thickness = -1
boxed_img = cv2.rectangle(img, start_point, end_point,
color, thickness)
if ss_readable_items > 0 and highlight_readable_text and not (
ss_matches > 0 and action in ("Highlight", "Redact")):
highlighted_img = boxed_img.copy()
# Highlight found matches of the search string
if ss_matches > 0 and action == "Highlight":
cv2.addWeighted(boxed_img, 0.4, highlighted_img, 1 - 0.4, 0,
highlighted_img)
# Redact found matches of the search string
elif ss_matches > 0 and action == "Redact":
highlighted_img = boxed_img.copy()
# cv2.addWeighted(boxed_img, 1, highlighted_img, 0, 0, highlighted_img)
# save the image
cv2.imwrite("highlighted-text-image.jpg", highlighted_img)
# Displays window showing readable text fields or the highlighted or redacted data
if show_comparison and (highlight_readable_text or action):
title = input_file if input_file else 'Compare'
conc_img = cv2.hconcat([initial_img, highlighted_img])
display_img(title, conc_img)
# Generates the text content of the image
output_data = None
if generate_output and details:
output_data = generate_ss_text(details)
# Prints a summary to the console
if input_file:
summary = {
"File": input_file,
"Total readable words": ss_readable_items,
"Total matches": ss_matches,
"Confidence score": ss_confidence
}
# Printing Summary
print(
"## Summary ########################################################"
)
print("\n".join("{}:{}".format(i, j) for i, j in summary.items()))
print(
"###################################################################"
)
return highlighted_img, ss_readable_items, ss_matches, ss_confidence, output_data
def ocr(self, psm: int = 4, oem: int = 1):
config = '--psm {} --oem {}'.format(psm, oem)
ocr_df = pytesseract.image_to_data(self.img,
config=config,
output_type=Output.DATAFRAME)
return TesseractOcrOperand(self, ocr_df)
def charsearchdata(what=49, graphic_enable=graphic_on):
data = pytesseract.image_to_data(gray, config='--psm 13')
print(data)
ap.add_argument("-i",
"--image",
required=True,
help="path to input image to be OCR'd")
ap.add_argument("-c",
"--min-conf",
type=int,
default=0,
help="mininum confidence value to filter weak text detection")
args = vars(ap.parse_args())
# load the input image, convert it from BGR to RGB channel ordering,
# and use Tesseract to localize each area of text in the input image
image = cv2.imread(args["image"])
rgb = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
results = pytesseract.image_to_data(rgb, output_type=Output.DICT, lang="jpn")
# loop over each of the individual text localizations
for i in range(0, len(results["text"])):
# extract the bounding box coordinates of the text region from
# the current result
x = results["left"][i]
y = results["top"][i]
w = results["width"][i]
h = results["height"][i]
# extract the OCR text itself along with the confidence of the
# text localization
text = results["text"][i]
conf = int(results["conf"][i])
return ''.join(cuit.groups())
#endregion
#region - "MAIN"
if __name__ == "__main__":
obtener_puntos()
niveles = sorted(niveles)
lineas = obtener_lineas(puntos, niveles)
obtener_secciones(lineas)
for sec in range(len(secciones)):
key = 'Seccion-' + str(sec + 1)
boxes = pytesseract.image_to_data(secciones[sec])
ocr_secciones[key] = limpiar_boxes(boxes)
'''cv2.imshow("Seccion", secciones[sec])
cv2.waitKey(0)'''
print(ocr_secciones)
cv2.destroyAllWindows()
#endregion
#region - PROCESAMIENTO PALABRAS
nombre_obtenido = False
for key in ocr_secciones:
if len(ocr_secciones[key]) > 3 and (not nombre_obtenido):
#print(ocr_secciones[key])
razon_social = obtener_razon_social(ocr_secciones[key])
import pytesseract
pytesseract.pytesseract.tesseract_cmd="C:/Program Files (x86)/Tesseract-OCR/tesseract.exe"
img=cv2.imread('1.jpg')
img1=img.copy()
img=cv2.cvtColor(img,cv2.COLOR_BGR2RGB)
#print(pytesseract.image_to_string(img ))
#for image to Boxes
imx,imy,imc=img.shape
print(imx)
boundary=pytesseract.image_to_boxes(img )
for b in boundary.splitlines():
b=b.split(' ')
x,y,w,h=int(b[1]),int(b[2]),int(b[3]),int(b[4])
cv2.rectangle(img,(x,imx-y),(w,imx-h),(255,0,0),1)
cv2.putText(img,b[0],(x,imx-y-20),cv2.FONT_HERSHEY_DUPLEX,0.34,(0,0,0),1)
cv2.imshow('IMAGE TO LETTERS', img)
#image to Data
boundary1=pytesseract.image_to_data(img1)
for c,b1 in enumerate(boundary1.splitlines()):
if(c!=0):
b1=b1.split()
if len(b1)==12:
x1,y1,w1,h1=int(b1[6]),int(b1[7]),int(b1[8]),int(b1[9])
cv2.rectangle(img1,(x1,y1),(w1+x1,h1+y1),(255,0,0),1)
cv2.putText(img1,b1[11],(x1,y1),cv2.FONT_HERSHEY_DUPLEX,0.34,(0,0,0),1)
cv2.imshow('Image TO DATA', img1)
cv2.waitKey(0)
contours, hierarchy = cv2.findContours(
dilation, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
im2 = img.copy()
file = open("recognized.json", "w+")
file.write("[")
file.close()
for idx, cnt in enumerate(contours):
x, y, w, h = cv2.boundingRect(cnt)
rect = cv2.rectangle(im2, (x, y), (x + w, y + h), (0, 255, 0), 2)
cropped = im2[y:y + h, x:x + w]
file = open("recognized.json", "a")
print("[INFO] reading text...")
text = pytesseract.image_to_data(cropped, output_type=Output.DICT)
print("[INFO] writing text...")
file.write(json.dumps(text))
if idx != len(contours) - 1:
file.write(",")
else:
file.write("]")
# print(text)
file.close()
print("[Finished] Data written to the file .\\recognised.json")
def grid_search_adaptive_binarization(
img_path,
cv_adaptive_sizes=[15, 25],
cv_adaptive_cs=[10],
cv_adaptive_methods=['ADAPTIVE_THRESH_MEAN_C'],
output=True,
eval_method='mlc',
ground_truth_path=None):
'''
try each size and C value to find best adaptive binarization params
eval_method = [mlc, wer]
evaluation measurement = WORD ERROR RATE
return dict of results
'''
if eval_method not in ['mlc', 'wer']:
print('error: invalid eval_method')
return False
if eval_method == 'wer' and ground_truth_path is not None:
# load
try:
ground_truth = open(ground_truth_path, 'r').read()
except:
print('error: cannot load ground truth from .txt')
if eval_method == 'wer' and ground_truth_path is None:
print(
'error: please provide a gound truth text if evaluation method == wer'
)
return False
import time
import numpy as np
print_time_once = True
gridsearch_result = []
for method in cv_adaptive_methods:
for c in cv_adaptive_cs:
for size in cv_adaptive_sizes:
start_time = time.time()
# apply treshold
img_bin_adapt = adaptive_binary(img_path, size, c, method)
# apply ocr
df_data = pytesseract.image_to_data(img_bin_adapt,
output_type='data.frame')
# measurement
if eval_method == 'mlc':
try:
length, text, measure = image_to_data_stats(
df_data, output=False)
except (e):
print('error: mean line confidence', e)
measure = np.nan
length = np.nan
elif eval_method == 'wer':
try:
hypothesis = ' '.join(
df_data[df_data['text'].notnull()].text.to_list())
length, _, _ = image_to_data_stats(df_data,
output=False)
measure = word_error_rate(ground_truth, hypothesis)
except:
measure = np.nan
length = np.nan
# add gs data
gridsearch_result.append({
'length': length,
eval_method: measure,
'size': size,
'c': c,
'method': method
})
# make the time stop iterat
time_end = round((time.time() - start_time), 2)
if print_time_once:
cv_iterations = len(cv_adaptive_sizes) * len(
cv_adaptive_cs) * len(cv_adaptive_methods)
print('eval_method:', eval_method, ' nr. of iterations:',
cv_iterations, 'est. time (min):',
round(cv_iterations * time_end / 60, 2))
print_time_once = False
if output:
print('GS>', 'method', method, 'size', size, 'c', c,
eval_method, measure, '\t time:', time_end)
plot_img(img_bin_adapt, 150)
plt.pause(0.05)
'''return dict and best index with best params'''
if eval_method == 'mlc':
best_idx, best_params = max(enumerate(gridsearch_result),
key=lambda item: item[1][eval_method])
elif eval_method == 'wer':
best_idx, best_params = min(enumerate(gridsearch_result),
key=lambda item: item[1][eval_method])
return gridsearch_result, best_idx, best_params
def img_to_data(img):
return pytesseract.image_to_data(img, output_type = pytesseract.Output.DICT)
def get_data(im):
return pytesseract.image_to_data(im, output_type=Output.DICT)
img = cv.imread(root_path + filename)
if img is None:
print("Error opening video stream or file")
sys.exit()
img_thresholding = img.copy()
import pytesseract
from pytesseract import Output
# custom_config = r'--oem 3 --psm 6'
custom_config = r'--oem 3 --psm 6 -l kor+kor_vert+eng' # korean + english
words_string = pytesseract.image_to_string(img)
words = pytesseract.image_to_data(img,
config=custom_config,
output_type=Output.DICT)
print(words.keys())
n_boxes = len(words['text'])
for i in range(n_boxes):
if int(words['conf'][i]) > 60:
(x, y, w, h) = (words['left'][i], words['top'][i], words['width'][i],
words['height'][i])
img = cv.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
cv.imshow('Resource', img)
#thresholding
def thresholding(image):
return cv.threshold(image, 0, 255, cv.THRESH_BINARY + cv.THRESH_OTSU)[1]
if count > 1000:
break
# cv2.imwrite("frame%d.jpg" % count, image) # save frame as JPEG file
success, image = vidcap.read()
# print('Read a new frame: ', success)
if (success):
count += 1
if (count > (13 * 30) and count % 50 == 0):
try:
img = image[650:702, 40:155]
img = (get_grayscale(img))
# Show Analysed Image
# cv2.imshow('img', img)
# cv2.waitKey(0)
d = pytesseract.image_to_data(img,
lang="eng",
config=custom_config,
output_type=Output.DICT)
print(d['text'])
n_boxes = len(d['text'])
for i in range(n_boxes):
matches = [
"REPLAY", "REPL", "REP", "PLAY", "EPL", "PLA"
]
if any(x in d['text'][i] for x in matches):
print("Goal scored detected at frame: {}".format(
count))
goalDetected = True
videosWithConfirmedGoal += 1
detectedGoalList.append([v, count])
# Overlay Bounding Boxes
# if int(d['conf'][i]) > 60:
def tsvOCR(img, savpath, tsvfile, p=1, append=True, psm=1, oem=3):
"""
Run OCR on an image and produce a TSV file along with text. Can limit which
files are added to the TSV by way of binomial sampling.
Arguments
--------------------------------------------------------------------------
img (str, PIL image) : Either the file path of an image or a PIL image object
savpath (str) : The file path to save the text output.
tsvfile (str) : Name of a TSV file that will be produced in the same
directory as the text output. Ignored if not specified.
p : Probability to use for binomial sampling. If 1, no sample
will be taken.
append (boolean) : If true, text output is appended to the file specified
by savpath. If the file does not exist, it will be created.
Note: this argument also applies to the TSV output if a
TSV file is specified.
psm (int) : Tesseract configuration for Page Segmentation Mode.
https://github.com/tesseract-ocr/tesseract/blob/master/doc/tesseract.1.asc
oem (int) : Tesseract configuration for OCR Engine mode.
https://github.com/tesseract-ocr/tesseract/blob/master/doc/tesseract.1.asc
Returns
--------------------------------------------------------------------------
none
"""
#if a filename is used for the image, load the image
if type(img) == str:
name = os.path.split(img)[1]
img = Image.open(img)
else:
name = img.info["name"]
# Set up tesseract config
if 0 <= psm <= 13:
pconfig = "--psm " + str(psm)
else:
return "Invalid tesseract config."
if 0 <= oem <= 3:
oconfig = "--oem " + str(oem)
else:
return "Invalid tesseract config."
tconfig = pconfig + " " + oconfig
#determine mode to save files
if append == True:
mode = {"mode": "a"}
else:
mode = {"mode": "w"}
#Get TSV data
tsvs = pytesseract.image_to_data(img, config=tconfig)
tdf = pandas.read_csv(StringIO(tsvs),
sep="\t",
engine="python",
error_bad_lines=False)
#Turn TSV data into text
par = 0
line = 0
text = ""
for row in tdf.itertuples():
token = str(row.text)
#adjust tokens
if token == "nan":
continue
if row.par_num != par:
par = row.par_num
line = row.line_num
token = "\n\n" + token + " "
elif row.line_num != line:
line = row.line_num
token = "\n" + token + " "
else:
token = token + " "
#compile text
text = text + token
#write text to file
ocrf = open(savpath, **mode)
ocrf.write(text)
ocrf.close()
#Drop TSV columns we don't need
tdf = tdf.drop(columns=[
"level", "page_num", "block_num", "par_num", "line_num", "word_num"
])
#Sample TSV data
if p != 1:
samp = random.binomial(size=tdf.count()[0], n=1, p=p)
r = 0
for s in samp:
if s == 0:
tdf = tdf.drop(r)
r += 1
#if we aren't left with an empty data frame
if tdf.count()[0] != 0:
#add a column for the image name
tdf["name"] = name
#determine if a header is needed
dirpath = os.path.split(savpath)[0]
if append == True and os.path.exists(
os.path.normpath(os.path.join(dirpath, tsvfile))) == True:
header = False
else:
header = True
#write data to TSV file
tsvf = open(os.path.normpath(os.path.join(dirpath, tsvfile)), **mode)
tdf.to_csv(tsvf, index=False, sep="\t", header=header)
tsvf.close()
return
def getPdfTextData():
# Scan the current directory for the .pdf file that will be converted
for file_ in os.scandir(os.getcwd()):
if file_.name.endswith('.pdf'):
filenames.append(file_.name)
# Load the .pdf file and save as .png
DPI = 600
pdf = Image(filename=filenames[0], resolution=DPI)
png_files = []
for page_num, page in enumerate(pdf.sequence):
page = Image(page)
page = page.convert('png')
temp_file_name = 'temp_' + str(page_num) + '.png'
png_files.append(temp_file_name)
page.save(filename=temp_file_name)
# Load the .png files and combine to make one image
img_arrays = []
#if len(png_files) > 1:
for file_ in (png_files):
# open the first image file
img = cv2.imread(file_)
sz = img.shape
start_row = round(0.05 * sz[0])
end_row = sz[0] - start_row
# convert the image to numpy array
img_arrays.append(np.array(img)) #[start_row:end_row,:,:]))
# delete the temporary png file that we just read
os.remove(file_)
final_img_arr = np.vstack(img_arrays)
img = PIL.Image.fromarray(final_img_arr).convert('L')
threshold = 200
img = img.point(lambda p: p > threshold and 255)
img.show()
# this is where we will temporarily store the extracted text
tsv_filename = "text.tsv"
# Get TSV information about the text and write the temp file
# --psm 3 or --psm 6 or --psm 11 is working well
custom_oem_psm_config = r'--psm 3'
text = pytesseract.image_to_data(img, config=custom_oem_psm_config)
file_object = open(tsv_filename, 'w')
file_object.write(text)
file_object.close()
# Read the TSV .txt file into a pandas dataframe
csv_table = pd.read_table(tsv_filename, sep='\t')
# delete the TSV (.txt) file we created above now that the data is read into pd
os.remove(tsv_filename)
# Drop the rows with NaN values
csv_table = csv_table.dropna()
csv_table = csv_table.reset_index(drop=True)
# Clean the image
# DOCUMENT TYPE 1
#Physician Referral - Print View https://prs.choosebroadspire.com/Template.aspx?PRSId=20191115... (Top)
#1 of 4 11/15/19, 2:22 PM (Always on Bottom)
# DOCUMENT TYPE 2
#11/8/2019 Physician Referral - Print View (Always on top)
#https://prs.choosebroadspire.com/Template.a... 1/2 (Always on bottom)
# Remove timestamps, datestamps, and website stamps
# Find and drop the lines of text that match regular expressions like:
# - reg_expr = 'http://%'
# - reg_expr = '[0-1][0-9]/[0-3][0-9]/[0-9][0-9], [0-12]+:[0-5][0-9] [AP]M'
print(csv_table)
# if document is of type 1
if csv_table.text.values[0] == 'Physician':
cond_line1 = csv_table.text.str.contains('https://.*', regex=True)
cond_line2 = csv_table.text.str.contains(
'[0-9]{1,2}/[0-9]{1,2}/[0-9]{1,2},', regex=True)
line1_tops = csv_table[cond_line1].top.values
line1_heights = csv_table[cond_line1].height.values
line2_tops = csv_table[cond_line2].top.values
line2_heights = csv_table[cond_line2].height.values
# now that we have the top of the bounding boxes for each line
# lets search through all other entries in the data frame that
# have tops in a small interval around the top of lines of interest
for top1, top2, height1, height2 in zip(line1_tops, line2_tops,
line1_heights, line2_heights):
cond11 = csv_table['top'].ge(top1 - height1 * 0.7)
cond12 = csv_table['top'].le(top1 + height1 * 0.7)
line1_table = csv_table[cond11 & cond12]
#get the indices for the table above and drop them from the final table
cond21 = csv_table['top'].ge(top2 - height2 * 0.7)
cond22 = csv_table['top'].le(top2 + height2 * 0.7)
line2_table = csv_table[cond21 & cond22]
line1_indices = line1_table.index.values
line2_indices = line2_table.index.values
csv_table = csv_table.drop(index=line1_indices)
csv_table = csv_table.drop(index=line2_indices)
csv_table = csv_table.reset_index(drop=True)
# if document is of type 2
if csv_table.head(1).text.str.contains('[0-9]{1,2}/[0-9]{1,2}/[0-9]{4}',
regex=True).values[0]:
# Because document of type 2 have header lines we want to remove containing dates,
# we need to be careful not to delete other lines within the document that contain dates.
# To do this, we find all strings that have the date form and filter those by adding the contstraint
# that the next index in the table should be Physician Referral
# get all a table with matches for dates
dates_cond = csv_table.text.str.contains(
'[0-9]{1,2}/[0-9]{1,2}/[0-9]{4}', regex=True)
dates_indices = csv_table[dates_cond].index.values
dates_indices1 = dates_indices.tolist()
line1_tops = []
line1_heights = []
for index in dates_indices:
if csv_table.at[index + 1, 'text'] == 'Physician' and csv_table.at[
index + 2, 'text'] == 'Referral':
line1_tops.append(csv_table.at[index, 'top'])
line1_heights.append(csv_table.at[index, 'height'])
else:
continue
cond_line2 = csv_table.text.str.contains('https://.*', regex=True)
line2_tops = csv_table[cond_line2].top.values
line2_heights = csv_table[cond_line2].height.values
# now that we have the top of the bounding boxes for each line
# lets search through all other entries in the data frame that
# have tops in a small interval around the top of lines of interest
print('********')
print(line1_tops)
print(line1_heights)
print(line2_tops)
print(line2_heights)
for top1, top2, height1, height2 in zip(line1_tops, line2_tops,
line1_heights, line2_heights):
print('****')
cond11 = csv_table['top'].ge(top1 - height1 * 0.7)
cond12 = csv_table['top'].le(top1 + height1 * 0.7)
line1_table = csv_table[cond11 & cond12]
line1_indices = line1_table.index.values
print(line1_indices)
csv_table = csv_table.drop(index=line1_indices)
csv_table = csv_table.reset_index(drop=True)
#get the indices for the table above and drop them from the final table
cond21 = csv_table['top'].ge(top2 - height2 * 0.7)
cond22 = csv_table['top'].le(top2 + height2 * 0.7)
line2_table = csv_table[cond21 & cond22]
line2_indices = line2_table.index.values
print(line2_indices)
csv_table = csv_table.drop(index=line2_indices)
csv_table = csv_table.reset_index(drop=True)
return csv_table
def check(path=r'./imgs/300/Sample11-0.jpg', dpi=300):
for p in range(11, 12):
for o in range(3, 4):
img = cv2.imread('./imgs/200/Sample21-0.jpg')
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
img1 = copy.deepcopy(img)
#temp = sys.stdout
#f = open('out.txt', 'w')
#sys.stdout = f
d = pytesseract.image_to_data(
img,
output_type=pytesseract.Output.DICT,
lang='eng',
config=
'-c tessedit_char_whitelist="$%@.,&():ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789 " --psm {} --oem {}'
.format(p, o))
n_boxes = len(d['level'])
print(d)
#f.close()
#sys.stdout = temp
print('Read image')
try:
width = img.shape[1]
height = img.shape[0]
fps = 1
font = cv2.FONT_HERSHEY_SIMPLEX
thickness = 2
color = (0, 0, 255)
fontscale = 1
org = (width - 1000, height - 100)
org1 = (width - 1000, height - 200)
#fourcc = VideoWriter_fourcc(*'XVID')
#video = VideoWriter('tess_with_sp_psm_{}_oem_{}.avi'.format(p, o), fourcc, float(fps), (width, height))
for i in range(n_boxes):
#if(d['width'][i] <= 10 or d['height'][i] <= 10):
# continue
if (d['text'][i] == ''):
continue
(x, y, w, h) = (d['left'][i], d['top'][i], d['width'][i],
d['height'][i])
cv2.rectangle(img, (x, y), (x + w, y + h), (0, 255, 0), 2)
img1 = img1[y - 5:y + h + 5, x - 5:x + w + 5]
print(
pytesseract.image_to_string(
img1,
output_type=pytesseract.Output.DICT,
lang='eng',
config=
'-c tessedit_char_whitelist="$%@.,&():ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789 " --psm {} --oem {}'
.format(p, o)))
cv2.imwrite('temp/{}.jpg'.format(i), img1)
img1 = copy.deepcopy(img)
text = 'w_{}_h_{}'.format(d['width'][i], d['height'][i])
image = cv2.putText(img1, text, org, font, fontscale,
color, thickness, cv2.LINE_AA)
image = cv2.putText(img1, d['text'][i], org1, font,
fontscale, color, thickness,
cv2.LINE_AA)
#video.write(img1)
cv2.imwrite(
'./{}_tess_with_sp_psm_{}_oem_{}.jpg'.format(dpi, p, o),
img)
print('image saved')
except Exception as e:
print('fsdfa')
print("{} {}".format(p, o))
print(e)
im_resized = im.resize(size, Image.ANTIALIAS)
temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.png')
temp_filename = temp_file.name
im_resized.save(temp_filename, dpi=(300, 300))
#pytesseract
#config = "-l eng --oem 1 --psm 6 -c preserve_interword_spaces=1x1 -c tessedit_char_whitelist=0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"
config = "-l eng --oem 1 --psm 6 -c tessedit_char_whitelist=0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"
text = pytesseract.image_to_string(temp_filename, config=config)
print(text)
rowtext.append(text)
#pytesseract score_retrieval
text_score = pytesseract.image_to_data(cleaned_img,
output_type='data.frame',
config=config)
text_score = text_score[text_score.conf != -1]
lines = text_score.groupby(['block_num'])['text'].apply(list)
conf = list(text_score.groupby(['block_num'])['conf'].mean())
conf_score_list.append(conf)
#CRNN_Sparse
#comment these two below lines for enabling pytesseract
# text = textdetector(temp_filename, model_crnnspar)
# print(text)
# rowtext.append(text)
csv_writer.writerow(rowtext)
row_para.append(rowtext)
csvfile.close()
from wand.image import Image as wand_Image
from PIL import Image as PIL_Image
from wand.color import Color
import os
import pytesseract
def build_images(force=False):
if not all([not force, os.path.isfile('./foo-0.png'), os.path.isfile('./foo-0.png')]):
all_pages = wand_Image(filename='./example.pdf', resolution=300)
for idx, page in enumerate(all_pages.sequence):
with Image(page) as i:
i.format = 'png'
i.background_color = Color('white')
i.alpha_channel = 'remove'
i.save(filename='foo-%s.png' % idx)
build_images()
boxes = pytesseract.image_to_boxes(PIL_Image.open('foo-0.png'), output_type=pytesseract.Output.DICT)
data = pytesseract.image_to_data(PIL_Image.open('foo-0.png'), output_type=pytesseract.Output.DICT)
osd = pytesseract.image_to_osd(PIL_Image.open('foo-0.png'), output_type=pytesseract.Output.DICT)