niftynet.layer.rand_rotation module

class RandomRotationLayer(name='random_rotation')

Bases: niftynet.layer.base_layer.RandomisedLayer

generate randomised rotation matrix for data augmentation

init_uniform_angle(rotation_angle=(-10.0, 10.0))

init_non_uniform_angle(rotation_angle_x, rotation_angle_y, rotation_angle_z)

randomise(spatial_rank=3)

layer_op(inputs, interp_orders, *args, **kwargs)