def test_imexport_imimport():
shape = (2, 2)
image = np.zeros(shape)
with expected_warnings(['precision loss']):
pil_image = ndarray_to_pil(image)
out = pil_to_ndarray(pil_image)
assert out.shape == shape
def __call__(self, img):
noise_type = random.random()
if noise_type < .33:
noise_type = 's&p'
elif noise_type < .66:
noise_type = 'speckle'
elif noise_type < 1.:
noise_type = 'gaussian'
else:
return img
nd_img = np.array(img) / 255.0
nd_noised = random_noise(np.asarray(nd_img),
mode=noise_type,
clip=True)
img = ndarray_to_pil(nd_noised)
return img
def roundtrip_pil_image(self, x):
pil_image = ndarray_to_pil(x)
y = pil_to_ndarray(pil_image)
return y
def test_imexport_imimport():
shape = (2, 2)
image = np.zeros(shape)
pil_image = ndarray_to_pil(image)
out = pil_to_ndarray(pil_image)
assert out.shape == shape
def test_imexport_imimport():
shape = (2, 2)
image = np.zeros(shape)
pil_image = ndarray_to_pil(image)
out = pil_to_ndarray(pil_image)
assert out.shape == shape
def roundtrip_pil_image(self, x):
pil_image = ndarray_to_pil(x)
y = pil_to_ndarray(pil_image)
return y
draw.text((10, 60),"This Homework is cooly cool",(255), font=font)
draw.text((10, 85),"This Homework is cooly cool",(255), font=font)
draw.text((10, 110),"This Homework is cooly cool",(255), font=font)
draw.text((10, 135),"This Homework is cooly cool",(255), font=font)
draw.text((10, 160),"This Homework is cooly cool",(255), font=font)
draw.text((10, 185),"This Homework is cooly cool",(255), font=font)
draw.text((10, 210),"This Homework is cooly cool",(255), font=font)
draw.text((10, 235),"This Homework is cooly cool",(255), font=font)
draw.text((10, 260),"This Homework is cooly cool",(255), font=font)
y_test_missing.save('y_test_missing.png')
#Building and saving the mask
mask = ImageChops.difference(y_test_missing, y_test)
mask = np.array(pil_to_ndarray(mask) <= 0.1, dtype=int)
mask_img = mask * 255
mask = ndarray_to_pil(mask)
mask_img = ndarray_to_pil(mask_img)
mask_img.save('mask.png', 'PNG')
psnr_y_test_y_test_missing = peak_signal_noise_ratio(pil_to_ndarray(y_test), pil_to_ndarray(y_test_missing))
ssim_y_test_y_test_missing = structural_similarity(pil_to_ndarray(y_test), pil_to_ndarray(y_test_missing))
print(psnr_y_test_y_test_missing)
print(ssim_y_test_y_test_missing)
print('\n')
"""
plt.imshow(y_train, cmap='gray')
plt.show()
plt.imshow(y_test, cmap='gray')
plt.show()