Note

This page is a reference documentation. It only explains the class signature, and not how to use it. Please refer to the user guide for the big picture.

nidl.volume.transforms.augmentation.RandomGaussianBlur¶

class nidl.volume.transforms.augmentation.RandomGaussianBlur(sigma=(0, 2), **kwargs)[source]¶

Bases: VolumeTransform

Blur a 3d volume using a Gaussian filter with random kernel size.

It handles a numpy.ndarray or torch.Tensor as input and returns a consistent output (same type and shape). Input shape must be $(C, H, W, D)$ or $(H, W, D)$ (spatial dimensions).

Parameters:

sigma: (float, float) or (float, float, float, float, float, float), default=(0, 2): Range of the standard deviation $\sigma$ of the Gaussian kernel applied to blur the volume. If two values $(a,b)$ are provided, then $\sigma \sim \mathcal{U}(a, b)$ . If six values $(a_1, b_1, a_2, b_2, a_3, b_3)$ are provided, then one standard deviation per spatial dimension is sampled $\sigma_i \sim \mathcal{U}(a_i, b_i)$ for $i=1,2,3$ .
kwargs: dict: Keyword arguments.

__init__(sigma=(0, 2), **kwargs)[source]¶

apply_transform(data)[source]¶

Blur the input with a Gaussian filter.

Parameters:

data: np.ndarray or torch.Tensor: Input volume with shape $(C, H, W, D)$ or $(H, W, D)$ . Standard deviations in the Gaussian filter are equal across channels.

Returns:

data: np.ndarray or torch.Tensor: Blurred volume. Output type and shape are the same as input.