def test_ray_project_random(self):
"""Tests that the end point of the ray projects at the good location."""
tensor_size = np.random.randint(3)
tensor_shape = np.random.randint(1, 10, size=(tensor_size)).tolist()
random_point_2d = np.random.normal(size=tensor_shape + [2])
random_focal = np.random.normal(size=tensor_shape + [2])
random_principal_point = np.random.normal(size=tensor_shape + [2])
ray_3d = perspective.ray(random_point_2d, random_focal,
random_principal_point)
point_2d = perspective.project(ray_3d, random_focal, random_principal_point)
self.assertAllClose(random_point_2d, point_2d, rtol=1e-3)
def test_unproject_project_random(self):
"""Tests that unprojecting and projecting gives and identity mapping."""
tensor_size = np.random.randint(3)
tensor_shape = np.random.randint(1, 10, size=(tensor_size)).tolist()
random_point_2d = np.random.normal(size=tensor_shape + [2])
random_focal = np.random.normal(size=tensor_shape + [2])
random_principal_point = np.random.normal(size=tensor_shape + [2])
random_depth = np.random.normal(size=tensor_shape + [1])
point_3d = perspective.unproject(random_point_2d, random_depth,
random_focal, random_principal_point)
point_2d = perspective.project(point_3d, random_focal,
random_principal_point)
self.assertAllClose(random_point_2d, point_2d, rtol=1e-3)
def test_project_ray_random(self):
"""Tests that that ray is pointing toward the correct location."""
tensor_size = np.random.randint(3)
tensor_shape = np.random.randint(1, 10, size=(tensor_size)).tolist()
random_point_3d = np.random.normal(size=tensor_shape + [3])
random_focal = np.random.normal(size=tensor_shape + [2])
random_principal_point = np.random.normal(size=tensor_shape + [2])
random_depth = np.expand_dims(random_point_3d[..., 2], axis=-1)
point_2d = perspective.project(random_point_3d, random_focal,
random_principal_point)
ray_3d = perspective.ray(point_2d, random_focal, random_principal_point)
ray_3d = random_depth * ray_3d
self.assertAllClose(random_point_3d, ray_3d, rtol=1e-3)
def generate_ground_image(height,
width,
focal,
principal_point,
camera_rotation_matrix,
camera_translation_vector,
ground_color=(0.43, 0.43, 0.8)):
"""Generate an image depicting only the ground."""
batch_size = camera_rotation_matrix.shape[0]
background_image = np.ones((batch_size, height, width, 1, 1),
dtype=np.float32)
background_image[:, -1, ...] = 0 # Zero the bottom line for proper sampling
# The projection of the ground depends on the top right corner (approximation)
plane_point_np = np.tile(np.array([[3.077984, 2.905388, 0.]],
dtype=np.float32), (batch_size, 1))
plane_point_rotated = rotation_matrix_3d.rotate(plane_point_np,
camera_rotation_matrix)
plane_point_translated = plane_point_rotated + camera_translation_vector
plane_point2d = \
perspective.project(plane_point_translated, focal, principal_point)
_, y = tf.split(plane_point2d, [1, 1], axis=-1)
sfactor = height/y
helper_matrix1 = np.tile(np.array([[[1, 0, 0],
[0, 0, 0],
[0, 0, 0]]]), (batch_size, 1, 1))
helper_matrix2 = np.tile(np.array([[[0, 0, 0],
[0, 1, 0],
[0, 0, 1]]]), (batch_size, 1, 1))
transformation_matrix = tf.multiply(tf.expand_dims(sfactor, -1),
helper_matrix1) + helper_matrix2
plane_points = grid.generate((0., 0., 0.),
(float(height), float(width), 0.),
(height, width, 1))
plane_points = tf.reshape(plane_points, [-1, 3])
transf_plane_points = tf.matmul(transformation_matrix,
plane_points,
transpose_b=True)
interpolated_points = \
trilinear.interpolate(background_image,
tf.linalg.matrix_transpose(transf_plane_points))
ground_alpha = (1- tf.reshape(interpolated_points,
[batch_size, height, width, 1]))
ground_image = tf.ones((batch_size, height, width, 3))*ground_color
return ground_image, ground_alpha
def generate_ground_image(height,
width,
focal,
principal_point,
camera_rotation_matrix,
camera_translation_vector,
ground_color=(0.43, 0.43, 0.8)):
"""Generate an image depicting only the ground."""
background_image = np.ones((height, width, 1, 1), dtype=np.float32)
background_image[-1,
...] = 0 # Set the bottom line to 0 for proper sampling
# The projection of the ground depends on the top right corner (approximation)
plane_point_np = np.array([[3.077984, 2.905388, 0.]], dtype=np.float32)
plane_point2d = \
perspective.project(rotation_matrix_3d.rotate(plane_point_np,
camera_rotation_matrix) +
camera_translation_vector.T, focal, principal_point)
_, y = tf.split(plane_point2d, [1, 1], axis=-1)
sfactor = y / 256.
transformation_matrix = (1/sfactor)*np.array([[1, 0, 0],
[0, 0, 0],
[0, 0, 0]]) + \
np.array([[0, 0, 0],
[0, 1, 0],
[0, 0, 1]])
plane_points = grid.generate(
(0., 0., 0.), (float(height), float(width), 0.), (height, width, 1))
plane_points = tf.reshape(plane_points, [-1, 3])
transf_plane_points = tf.matmul(transformation_matrix,
plane_points,
transpose_b=True)
interpolated_points = \
trilinear.interpolate(background_image, tf.transpose(transf_plane_points))
ground_alpha = (1 - tf.reshape(interpolated_points, [256, 256, 1]))
ground_image = tf.ones((256, 256, 3)) * ground_color
return ground_image, ground_alpha
