def test_normalize(self):
image = torch.tensor([0, 255, 255], dtype=torch.uint8)
hist, bin_centers = histogram(image,
source_range='dtype',
normalize=False)
expected = torch.zeros(256, dtype=torch.long)
expected[0] = 1
expected[-1] = 2
self.assertTrue(torch.equal(hist, expected))
hist, bin_centers = histogram(image,
source_range='dtype',
normalize=True)
expected = expected.float().div(3.0)
self.assertTrue(torch.equal(hist, expected))
def test_negative_image(self):
image = torch.tensor([-100, -1], dtype=torch.int8)
hist, bin_centers = histogram(image)
self.assertTrue(torch.equal(bin_centers, torch.arange(-100, 0)))
self.assertTrue(hist[0] == 1)
self.assertTrue(hist[-1] == 1)
self.assertTrue((hist[1:-1] == 0).all())
def test_peak_float_out_of_range_image(self):
image = torch.tensor([10, 100], dtype=torch.float)
hist, bin_centers = histogram(image, nbins=90)
# offset values by 0.5 for float...
self.assertTrue(
torch.equal(bin_centers,
torch.arange(10, 100).float().add(0.5)))
def test_peak_float_out_of_range_dtype(self):
image = torch.tensor([10, 100], dtype=torch.float)
hist, bin_centers = histogram(image, nbins=10, source_range='dtype')
self.assertTrue(
torch.allclose(torch.min(bin_centers), torch.tensor(-0.9)))
self.assertTrue(
torch.allclose(torch.max(bin_centers), torch.tensor(0.9)))
self.assertEqual(len(bin_centers), 10)
def test_peak_int_range_dtype(self):
image = torch.tensor([10, 100], dtype=torch.int8)
hist, bin_centers = histogram(image, source_range='dtype')
self.assertTrue(torch.equal(bin_centers, torch.arange(-128, 128)))
self.assertEqual(hist[128 + 10], 1)
self.assertEqual(hist[128 + 100], 1)
self.assertEqual(hist[128 + 101], 0)
self.assertEqual(hist.size(), torch.Size([256]))
def histogram_demo():
image = torch.tensor(data.camera())
f, (ax1, ax2) = plt.subplots(1, 2)
ax1.imshow(image)
ax1.set_title("original image")
ax1.get_xaxis().set_visible(False)
ax1.get_yaxis().set_visible(False)
hist, bins = histogram(image)
ax2.set_title("histogram of image")
ax2.hist(hist, bins=bins)
f.show()
def test_flat_int_range_dtype(self):
image = torch.linspace(-128, 128, 256).type(torch.int8)
hist, bin_centers = histogram(image, source_range='dtype')
self.assertTrue(torch.equal(bin_centers, torch.arange(-128, 128)))
self.assertEqual(hist.size(), torch.Size([256]))
def test_peak_uint_range_dtype(self):
image = torch.tensor([10, 100], dtype=torch.int8)
hist, bin_centers = histogram(image)
self.assertEqual(len(hist), len(bin_centers))
self.assertEqual(bin_centers[0], 10)
self.assertEqual(bin_centers[-1], 100)
def test_wrong_source_range(self):
image = torch.tensor([-1, 100], dtype=torch.int8)
with self.assertRaises(ValueError):
_, _ = histogram(image, source_range='foobar')
def test_input_tensor(self):
image = [10, 100]
with self.assertRaises(TypeError):
_, _ = histogram(image)