def test_digital_rating_on_half_zeros(self):
z = np.ones((1000, 500))
o = np.zeros((1000, 500))
img = np.hstack((z, o))
rate = quality_ratings.get_rate_by_name("digital")
rating = rate(img)
self.assertAlmostEqual(0.5, rating, 3)
def test_digital_rating_on_quater_zeros(self):
ones = np.ones((1000, 750))
zeros = np.zeros((1000, 250))
img = np.hstack((zeros, ones))
rate = quality_ratings.get_rate_by_name("digital")
rating = rate(img)
self.assertAlmostEqual(0.75, rating, 3)
def test_digital_rating_on_3d_quater_zeros(self):
img = np.ones((1000, 1000, 3))
img[:,:,2] = 0
img[:250,:,0:2] = 0
rate = quality_ratings.get_rate_by_name("digital")
rating = rate(img)
self.assertAlmostEqual(0.75, rating, 3)
def test_digital_rating_on_3d_white(self):
img = np.ones((1000, 1000, 3))
rate = quality_ratings.get_rate_by_name("digital")
rating = rate(img)
self.assertAlmostEqual(1, rating, 3)
def test_digital_rating_on_3d_black(self):
img = np.zeros((1000, 1000, 3))
rate = quality_ratings.get_rate_by_name("digital")
rating = rate(img)
self.assertAlmostEqual(0, rating, 3)
def test_digital_rating_on_white(self):
'''All pixels aren't black - full data - goo quality'''
img = np.ones((1000, 1000))
rate = quality_ratings.get_rate_by_name("digital")
rating = rate(img)
self.assertAlmostEqual(1, rating, 3)
def test_digital_rating_on_black(self):
'''All pixels are black - no data - bad quality'''
img = np.zeros((1000, 1000))
rate = quality_ratings.get_rate_by_name("digital")
rating = rate(img)
self.assertAlmostEqual(0, rating, 3)
def test_analog_rating_on_blank(self):
'''Image doesn't contain any noise - good quality'''
img = np.zeros((1000, 1000))
rate = quality_ratings.get_rate_by_name("analog")
rating = rate(img)
self.assertAlmostEqual(1, rating, 2)
def test_analog_rating_on_gaussian_noise_medium_sigma_floating_img(self):
img = np.random.normal(scale=12.7, size=(1000, 1000))
img = img.astype(float) / 255.
rate = quality_ratings.get_rate_by_name("analog")
rating = rate(img)
self.assertAlmostEqual(0.5, rating, 1)
def test_analog_rating_on_gaussian_noise_big_sigma(self):
img = np.random.normal(scale=20, size=(1000, 1000))
rate = quality_ratings.get_rate_by_name("analog")
rating = rate(img)
self.assertAlmostEqual(0.0, rating, 2)
def test_get_rate_by_name(self):
self.assertIsNotNone(quality_ratings.get_rate_by_name("analog"))
self.assertIsNotNone(quality_ratings.get_rate_by_name("digital"))