def execute_insertion(sql, values):
"""Runs an insertion query and returns the result
:param sql: sql query
:param value: values for the query
:returns: True
"""
with closing(mysql.connector.connect(**connetion_params)) as db:
with closing(db.cursor(dictionary=True, buffered=True)) as cursor:
cursor.execute(sql, values)
db.commit()
return True
def execute_selection(sql, values=None):
"""Runs a select query and returns the result
:param sql: sql query
:param value: values for the query
:returns: query result
"""
with closing(mysql.connector.connect(**connetion_params)) as db:
with closing(db.cursor(dictionary=True, buffered=True)) as cursor:
if values == None:
cursor.execute(sql)
else:
cursor.execute(sql, [values])
data = parse_result(cursor)
return data
def ref_region(
img: np.ndarray,
selem: Any = disk(5),
sigma: int = 3,
opening_se: np.ndarray = np.ones((10, 10)),
closing_se: np.ndarray = np.ones((5, 5)),
verbose: bool = False
):
"""
"""
# Perform histogram equalisation :
_img_eq = rank.equalize(img, selem=selem)
# Perform edge detection :
_edges = canny(_img_eq, sigma=3)
_filled = ndi.binary_fill_holes(_edges)
# Morphological processing :
_eroded = utils.closing(
utils.opening(np.float64(_filled), opening_se), closing_se
)
if verbose:
utils.side_by_side(img, _img_eq, title1="Original", title2="Histogram Equalised")
#plt.title('Lol')
utils.side_by_side(_img_eq, _filled, title1="Histogram Equalised", title2="Canny Edge Detection + Filled image")
#plt.title('Lal')
utils.side_by_side(_filled, _eroded, title1="Canny Edge Detection + Filled image", title2="Opening, closing")
#plt.title('Lel')
return _eroded
def closing_mask(m, img=None):
# we give some margin
marg = 200
aux = np.zeros((m.shape[0] + marg*2, m.shape[1] + marg*2))
aux[marg:marg+m.shape[0], marg:marg+m.shape[1]] = m
# relative to the shape of the mask to compute closing
val = min(m.shape)
size = (min(int(0.1 * val), 100), min(int(0.1*val), 100))
aux = closing(aux, size=size)
return aux[marg:marg+m.shape[0], marg:marg+m.shape[1]], None
def darkText(img):
"""
Generates the textboxes candidated based on BLACKHAT morphological filter.
Works well with dark text over bright background.
Parameters
----------
img : ndimage to process
Returns
-------
mask: uint8 mask with regions of interest (possible textbox candidates)
"""
kernel = np.ones((30, 30), np.uint8)
img_orig = cv2.morphologyEx(img, cv2.MORPH_BLACKHAT, kernel)
TH = 150
img_orig[(img_orig[:,:,0] < TH) | (img_orig[:,:,1] < TH) | (img_orig[:,:,2] < TH)] = (0,0,0)
img_orig = closing(img_orig, size=(1, int(img.shape[1] / 8)))
return (cv2.cvtColor(img_orig, cv2.COLOR_BGR2GRAY) != 0).astype(np.uint8)
def main():
img = cv2.imread('lena.bmp', 0)
# img is now a 512 x 512 numpy.ndarray
if not os.path.exists('Task-1'):
os.makedirs('Task-1')
if not os.path.exists('Task-2'):
os.makedirs('Task-2')
if not os.path.exists('Task-3'):
os.makedirs('Task-3')
if not os.path.exists('Task-4'):
os.makedirs('Task-4')
if not os.path.exists('Task-5'):
os.makedirs('Task-5')
# Generate and output an image with
# additive white Gaussian noise
# with amplitude = 10
img_gauss_10 = generate_Gaussian_noise(img, 0, 1, 10)
cv2.imwrite('Task-1/lena.gaussian.10.bmp', img_gauss_10)
# Generate and output an image with
# additive white Gaussian noise
# with amplitude = 30
img_gauss_30 = generate_Gaussian_noise(img, 0, 1, 30)
cv2.imwrite('Task-1/lena.gaussian.30.bmp', img_gauss_30)
# Generate and output an image with
# salt-and-pepper noise with
# threshold = 0.05
img_sp_05 = generate_salt_and_pepper_noise(img, 0, 1, 0.05)
cv2.imwrite('Task-2/lena.sp.05.bmp', img_sp_05)
# Generate and output an image with
# salt-and-pepper noise with
# threshold = 0.1
img_sp_10 = generate_salt_and_pepper_noise(img, 0, 1, 0.1)
cv2.imwrite('Task-2/lena.sp.10.bmp', img_sp_10)
# Run 3x3 box filter on the image
# with white Gaussian noise with
# amplitude = 10
img_gauss_10_box_3 = box_filter(img_gauss_10, 3)
cv2.imwrite('Task-3/lena.gaussian.10.box.3x3.bmp', img_gauss_10_box_3)
# Run 3x3 box filter on the image
# with white Gaussian noise with
# amplitude = 30
img_gauss_30_box_3 = box_filter(img_gauss_30, 3)
cv2.imwrite('Task-3/lena.gaussian.30.box.3x3.bmp', img_gauss_30_box_3)
# Run 3x3 box filter on the image
# with salt-and-pepper noise with
# threshold = 0.05
img_sp_05_box_3 = box_filter(img_sp_05, 3)
cv2.imwrite('Task-3/lena.sp.05.box.3x3.bmp', img_sp_05_box_3)
# Run 3x3 box filter on the image
# with salt-and-pepper noise with
# threshold = 0.1
img_sp_10_box_3 = box_filter(img_sp_10, 3)
cv2.imwrite('Task-3/lena.sp.10.box.3x3.bmp', img_sp_10_box_3)
# Run 5x5 box filter on the image
# with white Gaussian noise with
# amplitude = 10
img_gauss_10_box_5 = box_filter(img_gauss_10, 5)
cv2.imwrite('Task-3/lena.gaussian.10.box.5x5.bmp', img_gauss_10_box_5)
# Run 5x5 box filter on the image
# with white Gaussian noise with
# amplitude = 30
img_gauss_30_box_5 = box_filter(img_gauss_30, 5)
cv2.imwrite('Task-3/lena.gaussian.30.box.5x5.bmp', img_gauss_30_box_5)
# Run 5x5 box filter on the image
# with salt-and-pepper noise with
# threshold = 0.05
img_sp_05_box_5 = box_filter(img_sp_05, 5)
cv2.imwrite('Task-3/lena.sp.05.box.5x5.bmp', img_sp_05_box_5)
# Run 5x5 box filter on the image
# with salt-and-pepper noise with
# threshold = 0.1
img_sp_10_box_5 = box_filter(img_sp_10, 5)
cv2.imwrite('Task-3/lena.sp.10.box.5x5.bmp', img_sp_10_box_5)
# Run 3x3 median filter on the image
# with white Gaussian noise with
# amplitude = 10
img_gauss_10_med_3 = median_filter(img_gauss_10, 3)
cv2.imwrite('Task-4/lena.gaussian.10.median.3x3.bmp', img_gauss_10_med_3)
# Run 3x3 median filter on the image
# with white Gaussian noise with
# amplitude = 30
img_gauss_30_med_3 = median_filter(img_gauss_30, 3)
cv2.imwrite('Task-4/lena.gaussian.30.median.3x3.bmp', img_gauss_30_med_3)
# Run 3x3 median filter on the image
# with salt-and-pepper noise with
# threshold = 0.05
img_sp_05_med_3 = median_filter(img_sp_05, 3)
cv2.imwrite('Task-4/lena.sp.05.median.3x3.bmp', img_sp_05_med_3)
# Run 3x3 median filter on the image
# with salt-and-pepper noise with
# threshold = 0.1
img_sp_10_med_3 = median_filter(img_sp_10, 3)
cv2.imwrite('Task-4/lena.sp.10.median.3x3.bmp', img_sp_10_med_3)
# Run 5x5 median filter on the image
# with white Gaussian noise with
# amplitude = 10
img_gauss_10_med_5 = median_filter(img_gauss_10, 5)
cv2.imwrite('Task-4/lena.gaussian.10.median.5x5.bmp', img_gauss_10_med_5)
# Run 5x5 median filter on the image
# with white Gaussian noise with
# amplitude = 30
img_gauss_30_med_5 = median_filter(img_gauss_30, 5)
cv2.imwrite('Task-4/lena.gaussian.30.median.5x5.bmp', img_gauss_30_med_5)
# Run 5x5 median filter on the image
# with salt-and-pepper noise with
# threshold = 0.05
img_sp_05_med_5 = median_filter(img_sp_05, 5)
cv2.imwrite('Task-4/lena.sp.05.median.5x5.bmp', img_sp_05_med_5)
# Run 5x5 median filter on the image
# with salt-and-pepper noise with
# threshold = 0.1
img_sp_10_med_5 = median_filter(img_sp_10, 5)
cv2.imwrite('Task-4/lena.sp.10.median.5x5.bmp', img_sp_10_med_5)
# Use octagon as kernel and set the orgin is at the center
kernel = [
[-2, -1], [-2, 0], [-2, 1],
[-1, -2], [-1, -1], [-1, 0], [-1, 1], [-1, 2],
[0, -2], [0, -1], [0, 0], [0, 1], [0, 2],
[1, -2], [1, -1], [1, 0], [1, 1], [1, 2],
[2, -1], [2, 0], [2, 1]
]
# closing followed by opening on all noisy images
img_gauss_10_close_open = opening(closing(img_gauss_10, kernel), kernel)
img_gauss_30_close_open = opening(closing(img_gauss_30, kernel), kernel)
img_sp_05_close_open = opening(closing(img_sp_05, kernel), kernel)
img_sp_10_close_open = opening(closing(img_sp_10, kernel), kernel)
cv2.imwrite('Task-5/lena.gaussian.10.close.open.bmp', img_gauss_10_close_open)
cv2.imwrite('Task-5/lena.gaussian.30.close.open.bmp', img_gauss_30_close_open)
cv2.imwrite('Task-5/lena.sp.05.close.open.bmp', img_sp_05_close_open)
cv2.imwrite('Task-5/lena.sp.10.close.open.bmp', img_sp_10_close_open)
# opening followed by closing on all noisy images
img_gauss_10_open_close = closing(opening(img_gauss_10, kernel), kernel)
img_gauss_30_open_close = closing(opening(img_gauss_30, kernel), kernel)
img_sp_05_open_close = closing(opening(img_sp_05, kernel), kernel)
img_sp_10_open_close = closing(opening(img_sp_10, kernel), kernel)
cv2.imwrite('Task-5/lena.gaussian.10.open.close.bmp', img_gauss_10_open_close)
cv2.imwrite('Task-5/lena.gaussian.30.open.close.bmp', img_gauss_30_open_close)
cv2.imwrite('Task-5/lena.sp.05.open.close.bmp', img_sp_05_open_close)
cv2.imwrite('Task-5/lena.sp.10.open.close.bmp', img_sp_10_open_close)
def closing_mask(m):
# relative to the shape of the mask to compute closing
val = min(m.shape)
size = (min(int(0.1 * val), 100), min(int(0.1 * val), 100))
return closing(m, size=size)
kernel = np.ones((50, 2))
side_by_side(binaria,
cv.morphologyEx(binaria, cv.MORPH_OPEN, kernel),
title1='Original',
title2=f'cv.MORPH_OPEN with Kernel {kernel.shape}')
# As Gonzalez explained in the book, an opening is nothing but an erosion followed by a dilation.
# Our custom function ```opening(image, kernel)``` yields the same result as executing ```cv.morphologyEx(image, cv.MORPH_OPEN, kernel)```
# # Closing
# In[17]:
kernel = np.ones((10, 10))
side_by_side(binaria,
closing(binaria, kernel),
title1='Original',
title2=f'closing() with Kernel {kernel.shape}')
# In[18]:
kernel = np.ones((10, 10))
side_by_side(binaria,
cv.morphologyEx(binaria, cv.MORPH_CLOSE, kernel),
title1='Original',
title2=f'cv.MORPH_CLOSE with Kernel {kernel.shape}')
# In[19]:
kernel = np.ones((2, 50))
side_by_side(binaria,